Introduction

Biomass yield is an important component to consider in any program designed to derive energy from plant material. Plant size and architecture are important biomass yield parameters, and these parameters are often quite variable within a given species. Intraspecies variation in biomass yield can be a product of many factors. One such factor is the timing of the transition from vegetative to reproductive growth. The switch to flowering causes a diversion of resources from the continual production of photosynthetic material (leaves) to the terminal production of reproductive tissue (flowers, seeds, and fruit). Thus, one method to increase biomass yields is to extend the vegetative phase by delaying the transition to flowering.

The initiation of flowering during particular seasons can greatly enhance the reproductive success of different plant species. Plants have thus evolved mechanisms to sense seasonal environmental cues that serve as a signal to initiate the conversion from vegetative to reproductive growth. For example, many species flower in response to changes in photoperiod, the relative length of day and night (for review, see [1, 2]). Others require exposure to the prolonged cold of winter to promote flowering through a process known as vernalization (for review, see [3]).

In the model dicot Arabidopsis thaliana (hereafter referred to as Arabidopsis), the genetic basis of flowering time has been studied extensively, and a plethora of genes have been implicated in the control of flowering time, but the majority of flowering time variation in Arabidopsis is attributable to just a few loci. Molecular studies of flowering-time control in cereals such as wheat and barley demonstrate that there are some similarities, but also important differences, between monocots and dicots [4].

Genetic analysis of flowering in wheat and barley has led to the identification of three VERNALIZATION (VRN) loci that influence the flowering response to cold for these cool season grasses. VRN1 is a MADS-box transcription factor similar to the Arabidopsis genes APETALA1 and FRUITFUL [5]. Expression of VRN1 is induced by cold, and VRN1 represses the expression of VRN2. VRN2 encodes a small CCT-domain protein that acts to repress flowering. Allelic variation at VRN2 is often responsible for the difference in vernalization requirement in winter versus spring varieties [6]. VRN2 acts to repress VRN3, a homolog of the Arabidopsis FLOWERING LOCUS T (FT) gene, a strong promoter of flowering [7]. Photoperiod also strongly influences flowering time in the grasses, as demonstrated by studies in rice (Oryza sativa L.). In rice, the Arabidopsis FT homolog (Heading date 3a) is induced under permissive photoperiods and promotes flowering [8]. To date, FT appears to be universal floral promoter. In all plants where flowering-time genes have been investigated, orthologs of FT promote flowering [1].

Perennial grasses, such as Miscanthus (Miscanthus giganteus) and switchgrass (Panicum virgatum), are promising bioenergy crops because of the potential for high biomass yields with low inputs, traits which are key for economic feasibility and environmental sustainability. Switchgrass is a prairie grass native to and widespread in North America which is capable of growth on marginal agricultural lands (http://bioenergy.ornl.gov/papers/misc/switgrs.html). Determining which loci and the combinations of those loci that control the genetic basis of flowering time in perennial grasses could greatly facilitate the development of cultivars that produce maximal biomass for a given location. However, many of the perennial grasses being considered for dedicated bioenergy crops are difficult models for gene identification due to polyploidy, obligate out crossing, long life cycle, and large size.

In 2008, the Department of Energy Joint Genome Institute sequenced the genome of the grass Brachypodium distachyon (hereafter referred to as Brachypodium), with the objective of making this species a genetic model for grasses. Brachypodium has a number of physical attributes that make it suitable for a model organism, such as a small genome, diploid strains, short life cycle, transformability, and small size [9–11]. In addition to a complete genome sequence, additional genetic and molecular resources include an expressed sequence tag and bacterial artificial chromosome libraries, microarrays, collections of natural strains, and the development of mutant populations (http://www.brachypodium.org/). Brachypodium has also been shown to be syntenic with other grass genomes making it applicable for comparative genomics [12–14].

Brachypodium accessions are highly variable for flowering time and biomass production (Fig. 1), but a vernalization treatment reduces this variability by reducing the time needed for winter-annual accessions to flower [10, 11, 15, 16]. Information concerning the loci responsible for flowering-time differences in natural Brachypodium accessions is limited. Using the rice Hd3a (FT ortholog) sequence as a probe, it was demonstrated that in Brachypodium, FT exists as a single copy gene [13]. Transformation of different Brachypodium accessions with the floral repressor Terminal Flower 1 from Arabidopsis or perennial ryegrass (Lolium perenne L.) showed that both genes are functional in Brachypodium and cause a delay in flowering time [15, 17–19].

Fig. 1
figure 1

Variability in biomass accumulation among Brachypodium accessions. a Single plant from two different accessions, which flowered on the same date. The accession on the left (Bd1-1) required 5 months to flower and had more biomass compared to the accession on the right (Bd21), which required only 2 months to flower. b Two different accessions at 2 months, with the accession on the left (BdCzPP-2) growing faster and producing more biomass in a given time compared to the accession on the right (Bd21)

Full size image

In this study, we investigate the flowering behavior of natural accessions of Brachypodium in response to cold and photoperiod treatments. We report substantial variation in flowering time among Brachypodium accessions, which indicates potential for using natural variation to uncover genes that function in the regulation of flowering in grasses. We also present an initial characterization of putative VRN orthologs in Brachypodium.

Materials and Methods

Plant Materials

Brachypodium accessions used in this study are listed in Table 1. Seeds of Bd1-1, Bd2-3, Bd3-1, Bd18-1, and Bd21 were kindly provided by David Garvin (USDA-ARS, Plant Science Research Unit and Department of Agronomy and Plant Genetics, University of Minnesota, St. Paul, MN, USA). Seeds of BdCzPP-2, BdCzPP-3, DSNSCA-6, and RON-2 were collected in Spain in the summer of 2007 by A. Manzaneda. Seeds from diploid accessions collected in Turkey were kindly provided by Hikmet Budak (Sabanci University, Istanbul, Turkey) and John Vogel (USDA-ARS, Western Regional Research Center, Albany, CA, USA [20]).

Table 1 Accessions used in this study
Full size table

Plant Materials and Growth Conditions

Seeds of Bd1-1, Bd2-3, Bd3-1, Bd18-1, Bd21, BdCzPP-2, BdCzPP-3, DSNSCA-6, RON-2, and the Turkish accessions were imbibed overnight between wet paper towels in the dark at 4°C. Seeds were planted in Metro-Mix 360 (SUN GRO Horticulture) and grown for 4 weeks at 20°C with 20 h of light per day at a fluence rate of 60–70 μmol m−2 s−1. Seedlings were then moved to one of four conditions: (1) no vernalization, (2) 2 weeks of vernalization (not tested with Turkish accessions), (3) 4 weeks of vernalization, and (4) 6 weeks of vernalization. Seedlings were vernalized in a cold room at 4°C in short days (SD; 8:16 h L/D), with the exception of the Turkish accessions for which seedlings were vernalized under constant light conditions. Plants were subsequently placed into a growth chamber containing cool-white fluorescent bulbs (GE EcoLux F32T8) at 20°C in long days (LD; 20:4 h L/D) and flowering time scored. For the Turkish accession, phenotypic data represents five isolates, for example the phenotypic value for BdTR1 was determined by scoring the flowering time from 3 plants each of BrTR1a, BdTR1b, BdTR1c, BdTR1d, and BdTR1e. In photoperiod shift experiments, long days were 20:4 h L/D, and short days were 8:16 h L/D.

For studies of seed vernalization, seeds (n = 8) from accession Bd1-1 were surface sterilized by rinsing 2× in 70% EtOH, followed by two rinses in dH2O. Seeds were soaked in a flask with gentle swirling for 1 h prior to plating. Seeds were plated on agar-solidified medium as described previously [21] and placed at 4°C for 24 h. Seed were moved to one of four conditions: (1) no vernalization, (2) 2 weeks of vernalization, (3) 4 weeks of vernalization, and (4) 6 weeks of vernalization. Seeds were vernalized at 4°C with in short days (8:16 h L/D). After the treatment, seeds had no visible root emergence and were sowed in MM360 and placed at 20°C in long days (20:4 h L/D).

To test the effects of far-red enrichment on plant architecture and flowering time, 4-week-old plants were placed in long days (20:4 h L/D) supplemented with far-red LED bulbs. The ratio of red/far-red under these conditions is 0.15 compared to the unenriched conditions which have a ratio of 5.

Genotyping and Expression Analysis

Oligos were designed for polymerase chain reaction (PCR) that include the polymorphic region of intron1 of Bradi3g10010. Once amplified, PCR products were digested with MseI and run out on a 4% agarose gel. Digested DNA resulted in unique banding patterns for all four Bradi3g10010 classes.

RNA was prepared from two separate pools of 7-day-old seedlings (n = 3), harvested at 4 PM (corresponding to the time of peak VRN3/FT expression) using the Nucleospin RNAII kit (Clontech) according to the manufacturer’s instructions. cDNA synthesis was carried out using M-MLV reverse transcriptase (Promega) according to the manufacturer's instructions. Quantitative reverse transcription PCR (qRT-PCR) was performed with the 7000 Real Time PCR System (Applied Biosystems) using the DyNAmo™ Flash SYBR® Green qPCR Kit (Finnzymes). The PCR parameters were one cycle of 15 min at 95°C, 40 cycles of 15 s at 95°C, 20 s at 58°C, and 45 s at 72°C. The constitutively expressed putative Brachypodium homolog of UBIQUITIN-CONJUGATING ENZYME 18 (UBC18 Bradi4g00660) was used as a reference for gene expression as described [22]. PCR primer sequences are listed in Table 2. All results presented are an average of two biological replicates.

Table 2 Sequences of oligos used in this study
Full size table

Statistical Analysis

ANOVAs and p test were performed using Microsoft Excel. Least significant difference was calculated for each condition, using a weighted mean to correct for uneven sample size. Standard deviations were calculated using Microsoft Excel.

Results and Discussion

The Vernalization Response in Brachypodium Accessions

Flowering time was analyzed for 24 natural accessions of B. distachyon in two independent experiments (Table 1). Nine accessions isolated from various geographical locations were assayed for flowering time following 0, 2, 4, or 6 weeks of vernalization (Fig. 2), and 15 accessions collected from Turkey were similarly assayed after 0, 4, or 6 weeks of vernalization (Fig. 3). For these two independent experiments, young plants (3 weeks old) were vernalized. In both experiments, we observed variation between strains for flowering time in both vernalized and nonvernalized samples. However, vernalization led to a decrease in variation within and between accessions, and a decrease in the numbers of days required to flower for most accessions (Figs. 2 and 3). Some of the Turkish strains displayed variation within the line (data not shown), indicating that there may be flowering-time loci segregating in these lines (BdTR12 and BdTR9). These lines were not included in our analysis. In contrast to the other accessions, the Turkish accession contains multiple isolates and individual isolates are not genetically homogeneous [20, 32]. However, even for the genetically heterogenous Turkish accessions, an ANOVA of the data shows that the vast majority of the variability observed was between accessions. Three accessions (Bd21, Bd3-1, DSNSCA-6) behaved as spring annuals, in which the accession would flower without vernalization and a vernalization treatment did not greatly accelerate the flowering time (Fig. 2). Three accessions behaved as winter annuals (Bd1-1, Bd18-1, and RON-2), and these accessions are quite late flowering (or in our conditions do not flower) without a vernalization treatment, and vernalization of these accessions results in a large decrease in flowering time (Fig. 2). However, there is substantial variation within the spring and winter-annual classes, indicating that flowering might be controlled by multiple loci and/or alleles. Two polyploid accessions (BdCzPP-2 and BdCzPP-3) were intermediate for flowering time relative to the spring and winter annuals, and for these accessions, a vernalization treatment reduced the time to flower, suggesting that their growth habits are more similar to the winter-annual accessions.

Fig. 2
figure 2

Flowering time and vernalization responsiveness for nine Brachypodium accessions. Flowering time was determined after 0, 2, 4, or 6 weeks of a vernalization treatment as seedlings. Each time point consisted of three to six plants, and the experiment was terminated 120 days after the vernalization treatments. Accessions Bd21 and Bd21-3 exhibited identical profiles (data not shown); therefore, only Bd21 is included. Data points at 120 days represent plants that did not flower during the course of the experiment. Least significant differences for the four treatments are (α = 0.05); NV = 6.055, 2WK = 4.986, 4WK = 4.284, and 6WK = 3.558

Full size image
Fig. 3
figure 3

Flowering time and vernalization responsiveness for 15 Turkish Brachypodium accessions. Flowering time was determined for the accessions following 0, 4, or 6 weeks of vernalization as seedlings. Each time point consisted of six to 15 plants, and the experiment was terminated 120 days following the vernalization treatment. Columns without error bars (standard deviation) contained at least one plant that did not flower during the experiment

Full size image

The ability of certain species to undergo vernalization as an imbibed seed is common in cereals, whereas in other species, vernalization is only possible in plants that have reached a certain stage of maturity [23, 24]. Seed vernalization of wheat was used extensively in the last century to essentially convert winter varieties into hardy spring varieties, ensuring a stable food supply in regions of the world in which only winter varieties could survive spring plantings [25]. Because many cereals can be vernalized as seed, we wanted to determine if the same holds true for Brachypodium. We evaluated the vernalization response in imbibed seeds of a single accession (Bd1-1) in order to compare the response of vernalization to that seen in young plants (Fig. 4). Seed vernalization of Bd1-1 had the same promoting effect as in young plants demonstrating that imbibed seeds are also capable of responding to a vernalization treatment. Specifically, 2 weeks of cold treatment causes a slight decrease in flowering time, but the within variability is relatively high, similar to that seen in the nonvernalized plants. However, after 4 weeks of vernalization, the flowering time was reduced to less than half the time it took the nonvernalized plants to flower, and the variability is also greatly reduced. Six weeks of vernalization caused even earlier flowering, but the difference between 4 and 6 weeks (p = 0.053) was much less compared to the flowering time difference between 2 and 4 weeks (p = 0.003), indicating that the response in this accession is nearing saturation by 6 weeks.

Fig. 4
figure 4

Flowering time of accession Bd1-1 vernalized as seeds. Flowering time was determined for accession Bd1-1 after 0, 2, 4, or 6 weeks of vernalization as seed (see “Materials and methods” section). Each column represents eight plants, and error bars are the standard deviation

Full size image

Response to Photoperiod and Light Quality

In both wheat and barley, exposure of certain winter accessions to SD can hasten the transition to flowering in the absence of cold, i.e., SD can serve as a substitute for vernalization [26, 27]. Thus, we wanted to determine the effect of a SD treatment on flowering in Brachypodium accessions. Accordingly, four accessions with varying flowering times were germinated under long days. After 2 weeks of growth in LD, plants were given one of three treatments: (1) maintained in LD, (2) transferred to SD for 4 weeks and then returned to LD, or (3) transferred to and maintained in SD. All of the spring-annual accessions showed a similar delay in flowering when subjected to a SD shift; however, accession BdCzPP-2, which has a robust vernalization response, flowered earlier in response to the temporary SD shift (Fig. 5), whereas none of the accessions flowered when maintained in SD. Therefore, there appears to be variation in the efficacy of a temporary SD shift to promote flowering in Brachypodium, and we identified one accession in which SD substituted, at least partially, for vernalization. Future experiments will address the effects of SD on additional Brachypodium accessions as well as the optimal photoperiod length of the SD and the optimal length of time in SD required for the promotion of flowering.

Fig. 5
figure 5

A temporary photoperiod shift to SD accelerates flowering in winter-annual Brachypodium accessions. Plants were grown in LD for 2 weeks and were either (1) maintained in LD or (2) shifted to SD for 4 weeks and then returned to LD. The SD shift only accelerated flowering in the winter-annual accession. Days to flower represents the time for flowering to occur from when the seeds germinated

Full size image

In Arabidopsis, nearby vegetation (shade) causes a number of phenotypic changes, which has been collectively called the shade avoidance syndrome. These include plant elongation and an acceleration of flowering. Six Brachypodium accessions were tested under enriched far-red conditions, which mimic shade, and neither plant elongation nor accelerated flowering was observed (data not shown). Therefore, the specific R/FR ratio that triggers the shade avoidance syndrome in Arabidopsis does not have the same affect in Brachypodium, which is consistent with the fact that many grasses have evolved to grow at a high vegetation density.

Analysis of Putative Brachypodium VRN Orthologs

In wheat and barley, variation in the VRN genes (VRN1, VRN2, and VRN3) contributes to variation in flowering time [28, 29]. Brachypodium contains several genes with high relatedness to wheat and barley VRN genes. When comparing both the wheat and barley VRN2 and VRN3 protein sequences to the Brachypodium annotation, single genes stood out as being most similar: Bradi3g10010 (BdVRN2) and Bradi1g48830 (BdVRN3). A similar approach identified multiple genes in Brachypodium which exhibit relatedness to VRN1. This is not surprising given that VRN1 is a MADS-box transcription factor, a class of genes that exists as large gene families in flowering plants [30, 31]. In this work, we chose to characterize the two Brachypodium genes most similar to wheat/barley VRN1, Bradi1g08340 (BdVRN1L1) and Bradi1g59250 (BdVRN1L2).

We sequenced BdVRN2 from several accessions and identified two indel polymorphisms (24 and 35 bp) and a single nucleotide polymorphism in the first intron (Supp Figure 1). We genotyped all available accessions for BdVRN2 intron 1 polymorphisms and identified four genotypic classes (Table 3; Supplemental Figure 1). The three earliest accessions (Bd21, Bd21-3, and Bd3-1) comprised one genotypic class (class 1). These three accessions were unique in that a vernalization treatment did not greatly reduce flowering time. Another class (class 3) included most of the accessions analyzed in this study (19 out of 27), and 16 of these were spring annuals whereas three accessions behaved as winter annuals. The two diploid winter-annual accessions in this genotypic class, Bd18-1 and RON-2, require an extended vernalization treatment to initiate flowering (Fig. 2), indicating genetically that they may comprise a separate subgroup within this class where vernalization responsiveness and flowering time may be influenced by other loci (ex. VRN1 and VRN3). The two remaining genotypic classes (classes 2 and 4) were comprised of only winter-annual accessions (n = 5) and are part of a Brachypodium subgroup as determined by 143 SSR markers [32], where all the accessions are winter annuals. Therefore, there appears to be a correlation between different genotypes at BdVRN2 and flowering time, but it is not known if the polymorphisms affect BdVRN2 function or are just a product of population structure. Future experiments to evaluate the functionality of BdVRN2 alleles include expression studies and transgenic complementation.

Table 3 Genotype classes of the BdVRN2 intron (Supplemental Figure 1) in different Brachypodium accessions
Full size table

The expression level of putative Brachypodium VRN orthologs was determined by qRT-PCR. In wheat and barley, VRN2 is a repressor of flowering and is expressed at lower levels in spring accessions. BdVRN2 is also expressed at lower levels in the spring-annual accessions Bd21-3 and Bd3-1 (Fig. 6). Likewise, VRN3 is a promoter of flowering and the expression of BdVRN3 is elevated in these accessions. Expression levels of Brachypodium VRN1-like genes, however, did not completely correlate with flowering time. Expression of BdVRN1L1was elevated in the rapid-flowering Bd3-1 accession but was also elevated in the late flowering winter annuals Bd1-1 and Bd18-1. In the case of BdVRN1L2, expression levels were the highest in the late-flowering accessions BdCzPP-2 and BdCzPP-3, but it is important to note that these are polyploid accessions, and thus, it is unclear how many copies of BdVRN1L2 exist in these backgrounds and whether these copies are functional. Among the diploid accessions, the spring annuals Bd21-3 and Bd3-1 had the highest BdVRN1L2 expression levels, whereas expression was low in the winter annuals Bd1-1 and Bd18-1. The expression data indicate that the BdVRN1 genes may have different activity and roles in different accessions. Future experiments will examine changes in the expression of these genes in response to cold treatment and different photoperiod regimes.

Fig. 6
figure 6

Expression of Brachypodium putative VRN orthologs. Nonvernalized seedlings were assayed for VRN gene expression. Accession designations are shown on the x-axis and are ordered according to flowering time for the nonvernalized condition, while relative BdVRN gene expression (compared to an ubiquitin control) is shown on the y-axis

Full size image

Flowering time is an important trait to consider in the development of lignocellulosic biofuel crops as plants with delayed flowering will produce greater yields. Work done in cereals has identified several genes that regulate the transition to flowering (reviewed in [29, 33]), and these loci may prove to be valuable targets for the manipulation of flowering in biofuel crops. However, domesticated grasses such as wheat and barley may have lost some variability in flowering time through the process of domestication. Being a wild grass species, Brachypodium may serve as a useful model for the identification of novel loci involved in the control of flowering. Work presented here clearly demonstrates a large degree of variation in flowering time among natural accessions of Brachypodium.

Brachypodium is a genetically tractable system to test genes involved in flowering time and biomass from food and bioenergy crops, which are often polyploid and therefore difficult to manipulate genetically. The small size and easy growth requirements of Brachypodium allow for high through put genetic screens and gene discovery.