The reprogramming of gene expression appears as the major trend in synthetic and natural allopolyploids where expression of an important proportion of genes was shown to deviate from that of the parents or the average of the parents.
In this study, we analyzed gene expression changes in previously reported, highly stable synthetic wheat allohexaploids that combine the D genome of Aegilops tauschii and the AB genome extracted from the natural hexaploid wheat Triticum aestivum. A comprehensive genome-wide analysis of transcriptional changes using the Affymetrix GeneChip Wheat Genome Array was conducted.
Prevalence of gene expression additivity was observed where expression does not deviate from the average of the parents for 99.3% of 34 820 expressed transcripts. Moreover, nearly similar expression was observed (for 99.5% of genes) when comparing these synthetic and natural wheat allohexaploids. Such near-complete additivity has never been reported for other allopolyploids and, more interestingly, for other synthetic wheat allohexaploids that differ from the ones studied here by having the natural tetraploid Triticum turgidum as the AB genome progenitor.
Our study gave insights into the dynamics of additive gene expression in the highly stable wheat allohexaploids.
Polyploidy constitutes a common process of evolution in flowering plants (Wendel, 2000; Rieseberg & Willis, 2007; Doyle et al., 2008). Recent studies have shown that the high divergence between hybridizing species drives whole-genome duplication leading to allopolyploid formation (Buggs et al., 2009; Paun et al., 2009). Allopolyploidy is accompanied by important changes at the genetic, transcriptomic, epigenetic and proteomic levels (Chen, 2010; Flagel & Wendel, 2010; Buggs et al., 2011; Ha et al., 2011; Hu et al., 2011; Zhou et al., 2011). Such changes might also have major consequences for morphological characters, leading to phenotype novelty (Adams, 2007; Hegarty et al., 2009; Ramsey, 2011).
Consequences of allopolyploidy on gene expression have been widely studied in natural and synthetic allopolyploids. Most of the studies using microarray approaches relied on the comparison of the expression level between the allopolyploid and its parents and/or the average of its parents, expressed as the midparental value (MPV) (Hegarty et al., 2005; Wang et al., 2006; Gaeta et al., 2009; Pumphrey et al., 2009; Chagué et al., 2010; Flagel & Wendel, 2010). Strikingly, these studies have shown that an important proportion of genes is nonadditively expressed in allopolyploids where expression patterns deviate from the MPV. Gene expression changes in allopolyploids involve various mechanisms (Chen & Ni, 2006; Doyle et al., 2008; Hovav et al., 2008), but this ‘transcriptomic shock’, an extended view of the ‘genomic shock’ concept (McClintock, 1984), appears to be a common feature, even though it was evaluated using different allopolyploids and technical approaches.
Allohexaploid wheat (Triticum aestivum L., 2n = 6x = 42, AABBDD) has evolved through two recurrent allopolyploidization events and provides a good example of a relatively recent and genetically stable allopolyploid species (see, for review, Matsuoka, 2011). The first allopolyploidization event occurred < 0.5 million yr ago (mya) between T. urartu Tumanian ex Gandylian (2n = 2x = 14, AA) and an unidentified diploid Aegilops sp. of the section Sitopsis, close to Aegilops speltoides Tausch (2n = 2x = 14, SS), as B genome donor and gave rise to the tetraploid Triticum turgidum L. (2n = 4x = 28, AABB) (Riley & Chapman, 1958; Dvorak & Zhang, 1990). A second allopolyploidization occurred c. 10 000 yr ago between the early domesticated tetraploid T. turgidum ssp. dicoccum and the diploid goatgrass Aegilops tauschii Coss (2n = 2x = 14, DD) and gave rise to T. aestivum (2n = 6x = 42, AABBDD) (Kihara, 1944; McFadden & Sears, 1946; Huang et al., 2002).
Early studies employing newly synthesized wheat allopolyploids documented that the onset of both the allotetra- and the allohexapolyploidization events in wheat are associated with rapid and extensive structural changes such as DNA rearrangements (Feldman et al., 1997; Ozkan et al., 2001, 2003; Shaked et al., 2001; Kashkush et al., 2002, 2003; Han et al., 2005; Feldman & Levy, 2009). More recent and detailed studies do not support the idea that structural changes are important in synthetic wheat allohexaploids, where a high genetic stability of those combining the tetraploid AB genome and the diploid D genome has been observed (Mestiri et al., 2010; Zhao et al., 2011). The high stability of allohexaploid wheat is also confirmed by the possibility of extraction of viable tetraploid plants that have the AB genome of allohexaploid wheat (designated hereafter AhAhBhBh), through elimination of the D genome by recurrent backcrosses and cytogenetic characterization (Kerber, 1964; PFR8BC Mestiri et al., 2010).
By contrast, synthetic wheat allohexaploids were shown to exhibit homologous pairing irregularities and aneuploidy, the frequency of which depends on the genotype of parents (Mestiri et al., 2010). It was shown, for example, that those synthetic allohexaploids with the extracted tetraploid AhAhBhBh as AB genome progenitor were more stable and showed better regular homologous pairing and lower aneuploid frequency than those with the natural wheat allotetraploid T. turgidum (Mestiri et al., 2010). On the other hand, gene expression analysis of various synthetic wheat allohexaploids has shown an important proportion (3.5–18%) of nonadditively expressed genes, depending on the studies and the allohexaploid wheat material (Pumphrey et al., 2009; Akhunova et al., 2010; Chagué et al., 2010; Qi et al., 2012).
We report in this study the prevalence of gene expression additivity in the highly stable wheat allohexaploids with the extracted tetraploid AhAhBhBh as AB genome progenitor, and discuss the consequences and role of expression additivity in the stabilization of allopolyploids.
Materials and Methods
Plant materials, growth conditions and RNA samples
The two synthetic wheat allohexaploids, TC54 and TC36, used in this study were previously described by Mestiri et al. (2010). They were obtained by crossing the tetraploid ‘Tetra-Courtot’, extracted from the allohexaploid wheat T. aestivum cv Courtot, as the AB genome progenitor, with either AttD54 or AtsD36, accessions of A. tauschii as D genome progenitors (Fig. 1). Accession AttD54 belongs to the subspecies tauschii, whereas AtsD36 belongs to the subspecies strangulata. All plants were grown in the same conditions in growth chambers at 22°C and 16 h day-length. Total RNA was extracted from the shoots using the Plant RNeasy Mini kit (QIAGEN, Inc., Valencia, CA) according to the manufacturer's recommendations (Chagué et al., 2010).
Microarray hybridizations and statistical analyses
The analysis of gene expression changes was done on wheat shoot samples as previously described (Chagué et al., 2010). We used the available Affymetrix GeneChip Wheat Genome Array, on which 55 049 transcripts were represented by probesets. Statistical analyses were performed as in Chagué et al. (2010) using two biological replicates, according to the varmixt model (Delmar et al., 2005). A set of 34 820 transcripts considered as expressed in at least one of analyzed genotypes according to the Affymetrix intensity cutoff (I > 3) was considered for further comparisons. Gene expression in synthetic allohexaploids was evaluated by detecting nonadditively expressed transcripts with expression levels deviating from MPVs (P ≤ 5%), as well as those additively expressed with expression levels not deviating from MPV (P > 5%). Here, MPVs correspond to the hybridization intensities of mixtures of equal amounts of parental RNA, as detailed in Chagué et al. (2010).
Transcriptome divergence between the progenitors
Comparisons between the D genome progenitor AtsD36 and the AB genome progenitor ‘Tetra-Courtot’ showed 14.4% (4799) differentially expressed transcripts out of the 34 820 transcript set. Among these, c. 60% were more expressed in ‘Tetra-Courtot’ and 40% were more expressed in AtsD36 (Table 1). Similarly, transcripts differentially expressed between ‘Tetra-Courtot’ and the other D genome progenitor (AttD54) represent 13.1% (4578) of the considered probeset; c. 95% of these were also common with those divergent between ‘Tetra-Courtot’ and AtsD36 (Table 1).
Table 1. Differences in gene expression between progenitors, and additive and nonadditive expression observed in the wheat synthetic allohexaploids TC36 and TC54
AhBh = D
AhBh > D
AhBh < D
Both wheat synthetic allohexaploids have the extracted tetraploid genome AhBh ‘Tetra-Courtot’ as AB genome donor, whereas the Aegilops tauschii accession 54 (AttD54) and 36 (AtsD36) were the D genome donors of TC54 and TC36, respectively. Accession AttD54 belongs to the subspecies tauschii, whereas AtsD36 belongs to the subspecies strangulata.
Here, MPVs correspond to the hybridization intensities of mixtures of equal amounts of parental RNA, as detailed in Chagué et al. (2010).
Comparison between progenitors
AhBh vs AtsD36
AhBh vs AttD54
Common to (AhBh vs AtsD36) and (AhBh vs AttD54)
Regulation in synthetic allohexaploids
Additively expressed transcripts
Additively expressed transcripts in TC 36 (= MPV36)
Additively expressed transcripts in TC 54 (= MPV54)
Common to TC36 & TC54
Nonadditively expressed transcripts
Up-regulated in TC36 (> MPV36)
Up-regulated in TC54 (> MPV54)
Common to TC36 and TC54
Down-regulated in TC36 (< MPV36)
Down-regulated in TC54 (< MPV54)
Common to TC36 and TC54
Additive and nonadditive gene expression in synthetic wheat allohexaploids
Strikingly, additivity occurred for the majority of transcripts in the two analyzed synthetic wheat allohexaploids. Of the 34 820 transcripts, only 242 (0.7%) and 337 (0.9%) were nonadditively expressed in the synthetic allohexaploids TC36 and TC54, respectively (Table 1). An important proportion of nonadditively expressed transcripts was equally expressed between the progenitors of TC36 (118 transcripts, representing 48%) and TC54 (192 transcripts, representing 57%) synthetic wheat allohexaploids (Table 1). Additive transcripts are, in large majority (86%), equally expressed between the two parents.
Sixty-three nonadditively expressed transcripts (31 up- and 32 down-regulated) were common to both TC36 and TC54 allohexaploids, representing a nonrandom and a nonincidental distribution (Table 1). The cross-comparison showed that 37 of these (17 up- and 20 down-regulated) were similarly nonadditively expressed in at least one of the synthetic allohexaploids (JOY54 or JOY36) analyzed in a previous study (Chagué et al., 2010) (Supporting Information, Table S1). The later synthetic allohexaploids have the same A. tauschii accessions as D genome progenitors, whereas they differ from TC54 and TC36 studied here by having the natural wheat allotetraploid T. turgidum (cv Joyau) as the donor of the AB genome (Mestiri et al., 2010). For the remaining transcripts nonadditively expressed in one of the synthetic wheat allohexaploids but additively expressed in the other one (Table 1), we found that 49 transcripts from TC54 and 69 transcripts from TC36 were similarly nonadditively expressed in at least one of the previously analyzed synthetic allohexaploids JOY54 or JOY36 (Table S1).
In a previous investigation with JOY54 and JOY36, we validated, using semiquantitative reverse transcriptase PCR (RT-PCR), the robustness of μArray technique and the applied statistical method in revealing additively and nonadditively expressed transcripts (Chagué et al., 2010). As an ultimate check in the present study, we have also validated expression of eight transcripts using semiquantitative RT-PCR (Fig. S1). Two transcripts and one transcript, respectively, were confirmed as additively expressed in TC36 and TC54, whereas two and three other transcripts, respectively, were confirmed as nonadditively expressed (Fig. S1), exactly as also revealed by the μArray data analysis.
Functional categories of nonadditively expressed transcripts
Analysis of gene ontology (GO) terms of all nonadditively expressed transcripts, shared or nonshared between the two synthetic allohexaploids TC54 and TC36, is provided in Table S1. The expression status of these genes (additive or nonadditive) in the synthetic allohexaploids JOY36 and JOY54, analyzed previously (Chagué et al., 2010), is also presented (Table S1). Although it is not the objective of the present study, comparison of frequencies of occurrence biological process GO terms between additively and nonadditively expressed transcripts revealed that those involved in defense response are enriched in nonadditive down-regulated transcripts, whereas those involved in reproduction are enriched in nonadditive up-regulated transcripts synthetic (Table S2). As already observed for other extracted tetraploid plants (Kerber, 1964), some Tetra-Courtot plants exhibit necrotic leaves and reduced fertility in comparison to the natural allotetraploid T. turgidum (data not shown). It is possible that the down-regulation of transcripts related to defense and oxidative stress response is correlated with the fertility observed when the D genome is added to synthetic wheat allohexaploids TC36 and TC54.
Direct comparison between synthetics and with natural allohexaploids wheat
Direct comparison between TC54 and TC36 synthetic allohexaploids showed only 48 differentially expressed transcripts (data not shown). This low number confirmed the gene expression stability among synthetic wheat allohexaploids, irrespective of the variability of the progenitor of the D genome.
Comparisons between the two synthetic wheat allohexaploids TC54 and TC36 and the natural wheat allohexaploid cv Courtot showed, respectively, 186 (0.5%) and 143 (0.4%) differentially expressed transcripts (P ≤ 0.5) (Fig. 2). The proportion of differentially expressed transcripts in TC54 or TC36 compared with Courtot is about quarter that of those reported previously in other synthetic allohexaploids with the natural allotetraploid T. turgidum (cv Joyau) as donor of the AB genome (Chagué et al., 2010). Twenty-four and 58 of these differentially expressed transcripts were commonly up- and down-regulated, respectively, in both synthetic allohexaploids as compared with Courtot (Fig. 2). All the remaining transcripts, except one, were differentially expressed between Courtot and one of the synthetic allohexaploids, whereas they were equally expressed when comparing Courtot with the other synthetic allohexaploid (Fig. 2).
Fifty-one and 92 transcripts, respectively, were up- and down-regulated in TC36 as compared with the natural wheat allohexaploid Courtot (Fig. 2). Cross-comparison shows that 11 and 16 transcripts were nonadditive and up- and down-regulated, respectively, in TC36 as compared with both MPV36 and Courtot (Fig. 2). This proportion is significantly higher than the null hypothesis of random distribution H0 (χ2 = 1.42E-11). Similarly, 63 and 123 transcripts, respectively, were up- and down-regulated in TC54 as compared with the natural wheat allohexaploid cv Courtot (Fig. 2). Similar comparison shows that 31 and 15 of these were also nonadditive and up- and down-regulated, respectively, in TC54 as compared with both Courtot and MPV54 (Fig. 2), which is also higher than the null hypothesis of random distribution (χ2 = 8.44E–0.6).
As noticed in different studies (Wang et al., 2006; Pumphrey et al., 2009; Rapp et al., 2009; Chagué et al., 2010), microarray technologies such as the Affymetrix GeneChip Wheat Genome Array do not allow one to distinguish the expression level of each of the gene copies (homoeoalleles) and their respective contribution to the overall gene expression. This method may underestimate the number of nonadditively expressed genes, as we are not able to detect a situation where repression of one homoeoallele is compensated by the activation of the other one. Nevertheless, various studies of gene expression changes in different allopolyploid species have shown that the expression level in the allopolyploid is not just an average of that of the two parents, where a large proportion of nonadditively expressed genes were revealed (Hegarty et al., 2005; Tate et al., 2006; Wang et al., 2006; Gaeta et al., 2009; Chagué et al., 2010; Flagel & Wendel, 2010; Qi et al., 2012).
A large proportion of nonadditively expressed transcripts (c. 7%) was observed in other wheat synthetic allohexaploids that differ from the ones studied here by having the natural tetraploid species T. turgidum spp. durum cv Joyau as the AB genome donor (Chagué et al., 2010). Thus, one of the extraordinary finding of this study is that particular synthetic wheat allohexaploids, with the extracted tetraploid ‘Tetra-Courtot’ as AB genome progenitor, showed nearly complete gene expression additivity (> 99%), with 10 times fewer nonadditively expressed genes than those having the natural wheat allotetraploid T. turgidum. Our previous study also showed that the TC36 and TC54 synthetic wheat allohexaploids were more stable, with more complete homologous pairing and lower aneuploid frequencies than those with the AB genome of natural wheat allotetraploid (Mestiri et al., 2010). Altogether, the extracted tetraploid component ‘Tetra-Courtot’ appears more compatible for hybridization and allohexaploidization than the durum wheat T. turgidum.
Importantly, our results suggest that the allohexaploidization event per se might not lead to drastic changes in gene expression if the AB genome donor is similar to that of the natural allohexaploid wheat. The near-complete additive expression observed here also suggests transcriptome ‘coresidence’ between the tetraploid AB and the diploid D genome in these wheat allohexaploids, where the majority of gene expression is simply an average of that of their parents. Interestingly, as observed previously (Chagué et al., 2010), the genetic variability of the progenitor of the D genome (A. tauschii) does not lead to drastic changes in gene expression between synthetic wheat allohexaploids. In comparison with our previous study using other synthetic allohexaploids where a different AB genome progenitor has been used, our results also indicate that the AB genome component might contribute more to genetic and genomic stability of hexaploid wheat. As discussed earlier (Mestiri et al., 2010), it is difficult to know whether the AB genome of the natural wheat allohexaploid has been preselected upon the natural allohexaploidization, leading to highly stable wheat allohexaploids, or whether this stability is the result of modifications that occurred after the allopolyploidization event during the evolution of the natural allohexaploid. Either the preselection of the highly compatible BhBhAhAh genome upon natural allohexaploidization and/or its possible evolution and modification during c. 10 000 yr of coresidence with the D genome in common allohexaploid wheat might explain the predominance of gene expression additivity in TC54 and TC36 synthetic allohexaploids. It is therefore not possible to know whether this predominance reflects the gene expression trajectories of allohexaploid wheat evolution and thus the consequence of the evolutionary process of allohexaploidization, after c. 10 000 yr of coresidence between AABB and DD genomes; or is the result of preselection of the compatible AABB genome upon natural allohexaploidization. It is important, however, to note that only one or a few hybridization events led to the present-day natural wheat allohexaploids (Talbert et al., 1998). Moreover, the allohexaploid wheat is the only reported allopolyploid where it is possible to extract the genome of one of the progenitors, favoring more the hypothesis of a simple ‘coresidence’ between the progenitor genomes, where no drastic changes would have occurred, as we have shown here.
In conclusion, the present study highlights new insights into the dynamics of additive gene expression in the highly stable wheat allohexaploids.
Additive expression for the majority of genes, as observed here, may also provide a molecular basis for dosage balance and compensation (Birchler et al., 2005; Veitia et al., 2008), as well as interactions of parental gene copies with each other (Adams et al., 2003; Adams & Wendel, 2005; Jackson & Chen, 2010).
This project was supported by the ANR (Agence Nationale pour la Recherche)-Biodiversité project (ANR-05-BDIV-015). H.C., V.C. and D.A. were funded by an Evry-Genopole postdoctoral fellowship. C.S. was funded by a fellowship of the French Research Ministry (MENRT).