SALVAMENTO DE UN BIEN EN PELIGRO Artículo 240 Salvamento

In order to examine the relatedness of FT cluster sequences among the different accessions, A neighbour-joining phylogenetic tree (Fig 6.14) and a Median-joining haplotype network (Fig 6.13) were constructed.

A total of 32 haplotypes were identified by PopART. Overall, the haplotype network has a star-like shape (Fig 6.13), centred around one that represent the ancestral cultivated haplotype. Except for the haplotype representing C. reticulatum, which is more divergent, most of the other haplotypes differ by only a few mutations and are therefore variants of the ancestral. This pattern suggests that they originate recently and is indicative of a population expansion during the recent history of the species. The neighbor-joining tree obtained (Fig 6.14) show similar results; few of the accession were grouped with a threshold 50% of bootstrap support specified in the analysis. In both the neighbour-joining tree and haplotype network, the

Chapter 6 Exploring natural sequence variation within the FTa1-a2-c cluster

157 accessions show a lack of structure, as desi and kabuli accessions show no clear genetic distinction according to the FT genes, and only a weak tendency to group according to the geographic origin.

Figure 6.13 Haplotype network based on the entire FT cluster sequence of a chickpea collection and constructed

using the median-joining algorithm. Unique haplotype groups are represented as circles connected by hash marks that indicate base pair changes between haplotypes. Numbers on haplotypes indicates a representative accession of that haplotype (excluding the "ICC-" prefix). A list of all accessions sharing each haplotype can be found in appendix 6.5.

Chapter 6 Exploring natural sequence variation within the FTa1-a2-c cluster

158

Figure 6.14 Neighbour-joining phylogenetic tree based on the alignment of 96 DNA sequences from C.

reticulatum and C. arietinum. Sequences were aligned using ClustalW in Geneious 8 software. Numbers in branches indicates bootstrap support from 1000 replicates, with a minimum of 50% necessary to group accessions.

Chapter 6 Exploring natural sequence variation within the FTa1-a2-c cluster

159

6.4 Discussion

Cultivated chickpea exhibits an earlier flowering phenotype compared to its wild predecessor, and this is understood to result from of selective pressure exerted since the early Bronze age when chickpea underwent a transition from winter to summer cropping (Berger et al. 2004a). In chapter 4 of this thesis, two intraspecific population were used to map the source of this earliness to a region of chromosome 3 containing a cluster of FT genes. The expression patterns of chickpea FT genes, analysed in the same chapter, is consistent with their potential role in flowering promotion, as proven for phylogenetically close legume species (Laurie et al. 2011; Hecht et al. 2011). Differences observed in their expression between the parental accessions of the crosses suggest that early flowering of the domesticated lines is associated with earlier induction and higher expression levels of these genes, suggesting the possibility of cis-regulatory changes resulting in their de-repression. Here, we characterized sequence variation across the cluster as a first step in investigating the possible causes of this elevated expression.

Interspecific polymorphisms

Not surprisingly, a large number of polymorphisms were observed between the C.

reticulatum and C. arietinum parental lines in the ~55 kb region sequenced. The transcribed

regions of the three genes were well conserved, with only one non-synonymous mutation found in FTc second exon. Although this affected a highly conserved residue, no clear evidence could be found to link this change to a difference in flowering time or FT protein function. Also, both alleles appeared to be relatively common within the subset of domesticated lines examined, suggesting that the polymorphism arose after domestication. Most of the differences between the wild and cultivated FT sequences were found in the non- coding regions. The most striking sequence difference found in this study is a Ty1/copia type LTR-retrotransposon insertion in the third intron of CaFTa. This insertion was present in all cultivated alleles but not in the wild species, implying that the insertion may have taken place after the domestication process. However, due to the low number of wild accessions analysed here, this conclusion should be taken cautiously. Further sequencing of the cluster in a higher number of C. reticulatum accession is needed to confirm it. Such work is currently underway; one of the objectives of the Consultative Group for International Agricultural Research

Chapter 6 Exploring natural sequence variation within the FTa1-a2-c cluster

160 (CGIAR) Consortium Research Program 3.5 on Grain Legumes is to assess the molecular

diversity among C. reticulatum accessions held in their repositories

(http://chickpealab.ucdavis.edu/index.php/goals/cgiar-consortium-research-program-35/). The presence of transposable elements (TEs) can have a profound impact in the surrounding genes, a well-documented phenomenon that has been the subject of several recent reviews (Cui and Cao 2014; Cui and Cao 2015; Deng et al. 2016; Jern and Coffin 2008). In plants, LTR-transposons as that reported in the present study are an important source of gene evolution and, therefore, of plant variation and adaptation (Galindo-González et al. 2017); they can influence nearby genes by a wide range of diverse mechanisms, including altering splicing and polyadenylation patterns, affecting chromatin structure or by functioning as enhancers or promoters of gene transcription. The insertion of TEs in non-translated regions can also lead to aberrant splicing (Hirsch and Springer 2017), but no evidence was seen for different variants of FTa2 mRNA between the wild and domesticated chickpea accessions. There are several examples of how the insertion of a retrotransposon in the upstream region of a flowering-related gene can impair gene expression and affect flowering time (Bratzel and Turck 2015); In rice, a retrotransposon insertion in the upstream region of Hd1 (a rice CO homolog) causes a weaker expression pattern and a decreased photoperiod sensitivity of this variant allele comparted to wild type (Yano et al. 2000; Hori et al. 2016). In soybean, a

Ty1/copia-like retrotransposon inserted in the first intron of the E9 locus (a soybean FT

ortholog) attenuated its expression although no disruption of the mRNA was found (Lee and An 2015). The opposite effect, where retrotransposon insertions cause upregulation of the gene where is inserted, is also documented (Galindo-González et al. 2017). In a classic example, the increased apical dominance of maize compared to its wild ancestor teosinte is due to the enhanced expression of the gene TEOSINTE BRANCHED 1 (TB1), caused by retrotransposon insertion in the 5’ regulatory region (Studer et al. 2011). Similar reports can be found in the case of flowering genes; in wheat, a retrotransposon insertion in the promoter region of VRN3 (FT homolog) is associated with elevated expression of this gene and thus early flowering (Yan et al. 2006). In Medicago, retrotransposon insertions in either the last intron or in the 3’ region of FTa1 resulted in gain-of-function alleles conferring elevated

FTa1 expression and early flowering (Jaudal et al. 2013). These examples provide a potential

explanation for the constitutively high FTa2 expression in all cultivated lines reported in chapter 4. Since FTa2 is the gene were the insertion is localized, this gene would be the main affected, but changes in the chromatin and the recruitment of transcription factors needed to

Chapter 6 Exploring natural sequence variation within the FTa1-a2-c cluster

161 start transcription could be extended to surrounding genes and affect FTa1 in this case, due to its genomic proximity (less than 4 kbp).

Besides the transposon insertion, a high number of polymorphisms were found in the non- coding regions in the comparison of the FT cluster from PI489777 and ICC4958 accessions. Variation was particularly rich in the FTa1upstream region, where hundreds of SNPs and indels were detected between both alleles. In view of the evidence pointing toFTa1 as the key functional gene regulating flowering time in this cluster in other legumes (Laurie et al. 2011; Putterill et al. 2013; Hecht et al. 2011) and in chickpea in particular (Chapter 4), polymorphisms in the FTa1 promoter and introns might have particular relevance for regulation of its function. In Arabidopsis, the FT gene is tightly regulated by multiple transcription factors that targets specific sequences in regions around the gene and modulate its expression. Various proximal and distal regions in its promoter are crucial for proper regulation, such as two CO-responsive elements (CORE1 and 2), a distant CCAAT box (5.3 Kb upstream the ATG codon, recognized by Nuclear factor Y and involved in the stabilization of CO that is binded to CORE1/2) or a Dof-binding site for CYCLING DOF

FACTOR, among others (Tiwari et al. 2010; Adrian et al. 2010; Cao et al. 2014; Andrés and

Coupland 2012; Song et al. 2015). The distance between distal and proximal elements is determinant for its photoperiodic response; differences in the length of this region have no effect in LD. By contrast, deletions in this area result in elevated FT expression and early flowering under SD, while insertions have the opposite effect (Liu et al. 2014). In the same way, both the promoter and first intron contains a CArG box that is targeted by the flowering inhibitor FLC to repress FT (Helliwell et al. 2006). In legumes, variation in non-coding regions has also been linked to differences in FT expression. Polymorphisms in the UTR and intronic regions are associated with higher expression of an FT5a allele that causes early flowering in soybean (Takeshima et al. 2016). In narrow-leafed lupin, a deletion in the promoter region of an FTc homolog is associated with massively upregulated expression, and the loss of vernalization requirement for early (Nelson et al. 2017). Similar effect of cis- variation has also been reported in other flowering-related genes. For example, natural variation in the non-coding sequences of Arabidopsis FLC define five major haplotypes that display differential epigenetic silencing and expression levels of FLC, which in turn determine the degree of vernalisation responsiveness in accessions from different latitudes (Coustham et al. 2012; Li et al. 2014). A similar scenario has been described in Brassica

Chapter 6 Exploring natural sequence variation within the FTa1-a2-c cluster

162 effectiveness of the epigenetic silencing that vernalization exerts on this gene. In functional alleles, BoFLC.C2 stays repressed after vernalization, whereas in others the expression of this gene gradually recovers pre-vernalized values after transfer to warm, causing a late phenotype. Although the exact cause is still unknown, polymorphisms in the non-coding intronic regions, rather than those found in coding sequence, are more correlated with the differential vernalization response (Irwin et al. 2016).

Summarizing, a high number of polymorphisms between the wild and cultivated alleles of the

FT cluster were observed. Which ones among them could be involved in the observed

upregulation of these genes is still to be determined. Further research will be needed, analysing potential regulatory motifs present in the wild accessions and how the polymorphism presented in this study modify them in the cultivated germplasm.

Intraspecific polymorphisms

As expected in a crop species, the number of polymorphic sites was lower at the C. arietinum intraspecific level. Yet, almost 100 polymorphisms were detected among the 94 accessions analysed.

Similar to results obtained in the interspecific comparison, the coding sequences were highly conserved. Besides the substitution already discussed in the previous section, two more amino acid changes were found in the FTc protein. Only one of them, the change Arg130Met found in accession ICC4363, showed potential to confer early flowering in mutagenesis studies in Arabidopsis (Ho and Weigel 2014). Interestingly this accession was one of the earliest lines in both the field and phytotron under LD. However, its earliness might be due to another factor, so in order to confirm its influence on flowering, further screening of this mutation in the large chickpea collections available at research centres is needed to see its spread through the available germplasm and test whether they present or not association with any flowering time alteration.

The non-coding region accumulated most of the polymorphism. The most significant changes, because of their size and potential association with flowering time, were two big deletions (~30 kb each) and a ~750 bp insertion in the FTa1-FTa2 intergenic region. The deletions (type 1 and type 2) could be a source of variation for flowering time, as we found that chickpea lines with type 1 were always later to flower, whereas those with type 2 are later in LD but seems to be earlier in SD. Due to their big size, both deletions affect the complete

Chapter 6 Exploring natural sequence variation within the FTa1-a2-c cluster

163 genomic sequence of FTa2 and also the FTa1-FTa2 intergenic region. Each of these elements have the potential to cause alterations in flowering time, although FTa2 could be considered a less likely candidate for a number of reasons; first, the loss of this gene did not obviously affect any aspect of plant development or architecture, suggesting that perhaps

FTa2 is not functional or might be highly redundant with its paralog FTa1. The high

homology between FTa1 and FTa2 proteins and their tandem arrangement is shared in all galegoid legumes so far examined (Laurie et al. 2011; Hecht et al. 2011; Rajandran 2016) indicating that these genes have arisen through duplication in a common ancestor of this group. Second, Arabidopsis complementation studies realized in Medicago and pea suggest that this gene is the least effective for promotion of flowering among all FT examined (Laurie et al. 2011; Hecht et al. 2011).

By contrast, the region between FTa1 and FTa2 seems to be very important for the regulation of FTa1 expression; in Medicago, retrotransposon insertions in the orthologous region results in elevated FTa1 expression and early flowering as reported by Jaudal et al. (2013). A ~10 kb deletion in the same region has been associated with a major flowering QTL in lentil (Rajandran 2016). Finally, further polymorphism observed in this study also support the possible relevance of the FTa1-a2 intergenic region as a 750bp insertion was significantly associated with early flowering when plants were grown under SD.

Which part of the intergenic region is relevant in flowering regulation is unknown, but the intergenic region in chickpea is quite small, comprising only 3.6 Kbp. Most of it is occupied by a non-coding RNA (ncRNA) spanning 2.3 kbp, according to NCBI annotations (GeneID 105851807). Interestingly, both type 1 and type 2 eliminate the last portion of this ncRNA. The relevance of this ncRNA is not clear, but the role of such RNAs in gene regulation have gained notoriety in recent years, uncovering an enormous potential at the transcriptional, translational, and mRNA-stability level (Kaikkonen et al. 2011; Inouye 1988; Clancy 2008; Lau and Lai 2005). They can also act as a decoy for microRNAs (miRNAs), blocking their interaction with the authentic targets (Wu et al. 2013a). Interestingly, in the model species

Brachypodium distachyon there is evidence of miRNAs participating in the regulation of FT

genes (Wu et al. 2013b). More research is needed to define the extent and splicing pattern of this ncRNA, to determine which parts of the region are involved in the modulation of FTa1 expression and to investigate their mechanism(s) of action.

Chapter 6 Exploring natural sequence variation within the FTa1-a2-c cluster

164 A final question that arise from the results is that regarding the quite unique features observed in the case of chickpea FTa2. Despite an apparent lack of functionality, as discussed above, the features observed in this gene are at least worth a mention. In first place, its structure of seven exons and six introns is quite different to that of canonical FT genes, which consist of four exons and three introns. The transcription of the three additional exons in the 5’ UTR of this gene, annotated in the NCBI entry (GeneID 101496618), were confirmed here. Such structure has not been reported for any FT gene from legume species. Also, in the present chapter, we detected proliferation of distinct splice variants for FTa2. More than ten isoforms were identified from PCR using a single primer pair, and the use of different combinations could show an even higher number. Interestingly, half of these transcripts lacked the first coding exon, and would likely encode non-functional proteins; one fact that could possibly explain the apparent lack of function of this gene. There was no evidence for any alterations in the abundance of these alternative transcripts between wild and domesticated lines, suggesting that the RT1 insertion did not affect FTa2 splicing.

The question of the possible functional significance of these transcript variants remains open. Alternative splicing (AS) is an important gene regulatory mechanism that, among other functions, allows proteome diversification and can influence gene expression (Stamm et al. 2005), and evidence suggests that around 20% of plant genes show AS (Wang and Brendel 2006). Compared with the extensive knowledge of FT control at the transcriptional level, our understanding of FT regulation at the posttranscriptional level is limited, although several cases of alternative splicing of FT genes have been documented; two FT genes have been found in the tree London plane (Platanus acerifolia), that yielded more than 34 different transcript depending on the tissues and developmental states and suggesting multiple regulatory roles for the different isoforms (Zhang et al. 2011). In maize, there are 15 FT-like genes and two of them (ZCN18 and ZCN26) show mRNA splicing (Danilevskaya et al. 2008). Sunflower is another species in which FT alternative splicing have been also described, although this case looks more like an splicing defect resulting in intron retention rather than a regulated alternative splicing (Blackman et al. 2010). The grass species Brachypodium

distachyon possess 2 FT genes, FT1 and FT2, with redundant roles in induction of flowering.

Two isoforms of FT2 are formed by alternative splicing; FT2α (the fully functional isoform) and FT2β, which lacks a short section of the N-terminal domain and thus cannot form a functional florigen complex with FD and 14-3-3 proteins. FT1, FT2α and FT2β proteins can form heterodimers, but any complex containing the defective FT2β fails to promote flowering.

Chapter 6 Exploring natural sequence variation within the FTa1-a2-c cluster

165 The FT2β/FT2α heterodimer ratio progressively decreases during development, resulting in a gradual increase in the florigen activity level (Qin et al. 2017). Finally, one of the FT orthologues in Chrysanthemum morifolium is spliced in five different ways depending on developmental stage, with each variant showing different flowering-promotion capacity. One of them is lacking the exon 2 (analogue to those missing the first exon in this study) and was unable to promote flowering (Mao et al. 2016).

Future research analysing the expression of the different chickpea FTa2 isoforms in different developmental stages, tissues and environments will be necessary to fully understand their regulation and biological meaning. Due to the unknown effect of the retrotransposon insertion in the second intron in cultivated chickpea, C. reticulatum would be a preferable system in which to address these questions.

Genetic diversity among C. arietinum accessions

The nucleotide diversity found among C. arietinum accession analysed was very low, with values similar to those previously reported (Gujaria et al. 2011; Rajesh and Muehlbauer 2008; Jain et al. 2013). The low genetic diversity of chickpea is well documented, and is a consequence of the history of chickpea, which was domesticated around 10000 years ago during the Neolithic era. Since then, it suffered a series of four sequential bottlenecks during its evolution as a crop that highly constrained its genetic diversity, in contrast with other self-