•Common wheat is an allohexaploid species, derived through endoreduplication of an interspecific triploid hybrid produced from a cross between cultivated tetraploid wheat and the wild diploid relative Aegilops tauschii. Hybrid incompatibilities, including hybrid necrosis, have been observed in triploid wheat hybrids. A limited number of A. tauschii accessions show hybrid lethality in triploid hybrids crossed with tetraploid wheat as a result of developmental arrest at the early seedling stage, which is termed severe growth abortion (SGA). Despite the potential severity of this condition, the genetic mechanisms underlying SGA are not well understood.
•Here, we conducted comparative analyses of gene expression profiles in crown tissues to characterize developmental arrest in triploid hybrids displaying SGA.
•A number of defense-related genes were highly up-regulated, whereas many transcription factor genes, such as the KNOTTED1-type homeobox gene, which function in shoot apical meristem (SAM) and leaf primordia, were down-regulated in the crown tissues of SGA plants. Transcript accumulation levels of cell cycle-related genes were also markedly reduced in SGA plants, and no histone H4-expressing cells were observed in the SAM of SGA hybrid plants.
•Our findings demonstrate that SGA shows unique features among other types of abnormal growth phenotypes, such as type II and III necrosis.
Allopolyploid speciation, which involves the full duplication of a hybrid genome (Rieseberg & Willis, 2007), can result through various processes, including interspecific hybrid formation followed by endoreduplication. Common wheat (Triticum aestivum L.) is an allohexaploid species (BBAADD genome) that originated through allohexaploidization between cultivated tetraploid wheat Triticum turgidum L. (BBAA genome) and the wild diploid relative Aegilops tauschii Coss. (DD genome) (Goncharov, 2011). Hybrid breakdown between the species of wheat relatives inhibits allopolyploid speciation, a phenomenon that was first reported half a century ago in triploid wheat hybrids with a BA × D genome (Nishikawa, 1960, 1962a,b). More recently, the cv Langdon (Ldn) of T. turgidum subspecies durum was found to be an efficient BA genome parent for the production of hexaploid wheat synthetics (Matsuoka & Nasuda, 2004). This finding allowed us to produce a number of hexaploid synthetic wheat lines with the D genome derived from various A. tauschii accessions (Takumi et al., 2009; Kajimura et al., 2011). However, a number of abnormal phenotypes, including hybrid sterility and abnormal growth, have been observed in many F1 triploid plants of tetraploid wheat and A. tauschii (Nishikawa, 1960, 1962a,b; Matsuoka et al., 2007). Four types of abnormal growth phenotypes have been observed in triploid hybrids of Ldn and A. tauschii: hybrid necrosis types II and III, hybrid chlorosis, and severe growth abortion (SGA) (Mizuno et al., 2010). As the appearance of abnormal phenotypes in triploid wheat hybrids may function as a postzygotic hybridization barrier between BA and D genomes, and because they may prevent the formation of common wheat, it is important to understand the mechanisms underlying these processes.
The postzygotic hybridization barriers that exist between divergent lineages within the same species or between closely related species can accelerate the establishment of new diploid species (Bomblies & Weigel, 2007). The Dobzhansky–Muller (DM) model explains the evolutionary process underlying the generation of postzygotic hybridization barriers (Bomblies & Weigel, 2007; Rieseberg & Willis, 2007; Rieseberg & Blackman, 2010). Recent studies in Arabidopsis and lettuce have shown that hybrid necrosis, a type of hybrid breakdown, occurs as a result of programmed cell death triggered by autoimmune responses, and that the causal genes are defense response-related genes (Bomblies et al., 2007; Alcázar et al., 2009, 2010; Jeuken et al., 2009). In contrast to diploid species, hybrid barriers between related species negatively affect allopolyploid speciation. In hybrid necrosis of triploid wheat plants, autoimmune responses occur as a result of epistatic interactions between genes from BA and D genomes (Mizuno et al., 2010, 2011). Therefore, the incompatibility in triploid hybrids is considered to be a DM-type hybrid barrier.
In intraspecific crosses of common wheat cultivars, Ne1 and Ne2 loci are known to control hybrid necrosis (Tsunewaki, 1970). The Ne1 and Ne2 complementary genes are located on chromosome arms 5BL and 2BS, respectively (Chu et al., 2006). In type III necrosis of interspecific wheat hybrids, cell death occurs gradually in older tissues, while necrotic symptoms associated with type II necrosis only appear under low-temperature conditions (Mizuno et al., 2010). In addition to necrotic symptoms, low temperature induces repression of mitotic cell division in the shoot apices of hybrid plants displaying type II necrosis (Mizuno et al., 2011). Unlike type II and III hybrid necrosis, there is little information concerning the genetic and molecular basis of hybrid chlorosis and SGA in triploid wheat hybrids. With the exception of SGA, the abnormal growth phenotypes in triploid hybrids are transmitted to selfed progeny. Only SGA is completely lethal, because SGA-exhibiting hybrid plants cease development and growth after expansion of the second or third leaves (Nishikawa, 1960; Mizuno et al., 2010). Similar to type II necrosis under low-temperature conditions, it is possible that shoot apices of SGA triploids are abnormal.
Here, we aimed to elucidate the mechanisms inducing SGA in triploid wheat hybrids at the transcriptional and physiological levels. Based on our results, we discuss differences in the abnormal phenotype-induction mechanisms of SGA as well as other forms of hybrid necrosis reported previously.
Materials and Methods
In our previous study, tetraploid wheat accession Triticum turgidum ssp. durum cv Langdon (Ldn) was used as the female parent and was crossed with 122 A. tauschii Coss. accessions to artificially produce triploid wheat hybrids (Mizuno et al., 2010). The passport data of the A. tauschii accessions, including the geographical coordinates of the original collection sites, were reported previously (Matsuoka et al., 2007, 2008). In this study, we used the 10 triploid wheat hybrids listed in Table 1. Two tetraploid wheat accessions, T. turgidum ssp. dicoccoidesKU-8736A and T. turgidum ssp. carthlicumKU-138, were also used as female parents to produce several new triploid hybrids. The F1 triploid hybrids were grown individually in pots in a glasshouse at Kobe University (Kobe, Japan) to determine growth phenotypes. The triploid hybrids exhibiting normal growth features were classified as wildtype (WT). The temperature of the glasshouse was not regulated.
Table 1. Aegilops tauschii accessions used in this study for the production of triploid wheat hybrids
A. tauschii accession no.
Phenotype in triploid hybrids
Experiments in the present study
RT-PCR, reverse transcriptase-PCR; SGA, severe growth abortion; TEM, transmission electron microscopy; WT, wildtype.
KU, Plant Germ-Plasm Institute, Faculty of Agriculture, Kyoto University, Japan; PI, National Small Grains Research Facility, USDA-ARS, USA; IG, International Center for Agricultural Research in the Dry Areas (ICARDA); CGN, Centre for Genetic Resources, the Netherlands.
F1 hybrid seedlings of Ldn and PI476874 (WT) and those of Ldn and IG47182 (SGA) grown at 23°C for 3 wk were used for transmission electron microscopy (TEM). The proximal and distal parts of leaf blades were cut into 1 mm2 pieces and then incubated in a freshly prepared solution of 5 mM CeCl3 and 50 mM 3-(N-morpholino)propanesulfonic acid at pH 7.2 for 1 h. The incubated leaf samples were then pre-fixed in 0.1 M cacodylate (CAB) buffer (pH 7.2) containing 2.5% (v/v) glutaraldehyde and 2% (v/v) paraformaldehyde. After washing three times with CAB buffer, the leaf samples were fixed in CAB buffer supplemented with 1% (w/v) osmium tetroxide for 1 h, and then dehydrated and embedded in Epon 812 resin (Nisshin EM, Tokyo, Japan). The sample blocks were cut with a diamond knife (Diatome, Bienne, Switzerland) on a Reichert-Nissei Ultracut microtome (Leica AG, Vienna, Austria) to obtain ultrathin sections (90 nm), which were collected on copper grids (200 mesh). Sections were examined and photographed using a Hitachi-7100 transmission electron microscope (Hitachi, Tokyo, Japan) at an accelerating voltage of 75 kV. Three independently produced sections derived from two plants were used for the determination of survival rate. More than 90 mesophyll cells in a section were counted for the detection of dead cells, which were identified as cells in which the collapse of organelles, plasmolysis, and disruption of cell membranes had occurred.
Two interspecifically crossed progenies, F1 hybrids of Ldn and PI476874 for WT, and F1 hybrids of Ldn and IG47182 for SGA, were used for microarray analysis. Total RNA was extracted from the crown tissues of WT and SGA triploids grown at 23°C for 3 wk using an RNeasy Plant Mini kit (Qiagen). A KOMUGI 38k oligonucleotide DNA microarray (Agilent Technologies, Santa Clara, CA, USA) was supplied by the National BioResource Project (NBRP)-Wheat, Japan (http://www.shigen.nig.ac.jp/wheat/komugi) for the microarray analysis. Detailed information on the 38k microarray platform can be found in Kawaura et al. (2008) and the Gene Expression Omnibus (GEO) database of the National Center of Biotechnology Information (NCBI) website under GPL9805. Hybridization of Cy3-labeled cRNA to the microarrays and washing were performed using a Gene Expression Hybridization kit and Gene Expression Wash Pack (Agilent Technologies). Scanned images were analyzed with Feature Extraction Software 9.5 (Agilent Technologies) using default parameters to obtain background-subtracted and spatially detrended processed signal intensities. Two independent experiments were conducted for each sample. All microarray data were deposited as GSE33357 in the NCBI GEO database (http://www.ncbi.nlm.nih.gov/geo/), including supplementary files, GSM825110–GSM825113.
The functions of probes and genes were predicted by both BLAST and BLASTx searches (E value < 1E−10) against the DNA Data Bank of Japan (DDBJ) database. The NBRP KOMUGI website (http://www.shigen.nig.ac.jp/wheat/komugi/array/index.jsp) was also used for the functional identification of the proteins encoded by the probes. Protein Knowledgebase (UniProtKB; http://www.uniprot.org/) was used for the functional categorization of identified genes.
Reverse transcriptase (RT)-PCR analyses
Reverse transcriptase-PCR and quantitative RT-PCR analyses were performed using RNA isolated from crown tissues of the four WT (F1 from Ldn × PI476874, Ldn × KU-2022, Ldn × KU-2059, and Ldn × KU-2069), one type II necrosis (F1 from Ldn × KU-2025), and three SGA (F1 from Ldn × IG47182, Ldn × IG47188 and Ldn × IG120866) triploids (Table 1). Each plant was grown at 23°C for 3 wk, and total RNA was extracted from crown tissues using Sepasol-RNA I (Nacalai Tesque, Kyoto, Japan). First-strand cDNA was synthesized from DNase I-treated RNA samples using ReverTra Ace Reverse Transcriptase (Toyobo, Osaka, Japan) and an oligo(dT)20 primer. The gene-specific primer sets for RT-PCR are listed in Supporting Information (Table S1). PCR-amplified products were separated by electrophoresis on a 1.5% agarose gel and stained with ethidium bromide. The ubiquitin gene was used as an internal control.
The transcript accumulation of each gene was detected by quantitative RT-PCR using a LightCycler 480 Real-Time PCR System (Roche Diagnostics) with THUNDERBIRD SYBR qPCR Mix (Toyobo) and the gene-specific primer sets. The actin gene was used as an internal control, and the relative expression level was calculated as }, representing the value relative to the transcript abundances in the crown of WT (F1 plants of Ldn and PI476874).
A portable JUNIOR-PAM fluorometer (Heinz Walz GmbH, Effeltrich, Germany) was used for measurement of the maximum photochemical quantum yield of photosystem II (Fv/Fm). The first leaves of 3-wk-old WT and SGA hybrid plants (Table 1) were incubated under dark conditions for 1 h, and Fv/Fm was measured at 10 independent locations on each leaf blade according to the instructions in the operating manual for the fluorometer.
In situ hybridization analysis
In situ hybridization was performed as described previously (Shitsukawa et al., 2009). Briefly, shoot apices of triploid hybrids from crosses of Ldn and PI476874 (WT), and Ldn and IG47182 (SGA), were fixed in formalin/acetic acid/alcohol (FAA) solution (3.7% paraformaldehyde and 5% acetic acid) at 4°C overnight, dehydrated, and embedded in Paraplast resin (Oxford Labware, Mansfield, MA, USA) before being sliced into 10 μm sections. Hybridization was performed overnight at 52°C using a digoxigenin (DIG)-labeled RNA probe that was generated by in vitro transcription of a cDNA clone containing the wheat HistoneH4 gene (accession number X00043). After hybridization, labeled sections were washed twice with 0.5× SSC (7.5 mM sodium citrate and 75mM NaCl) at 52°C and treated with RNase. Signal detection and staining were performed according to Shitsukawa et al. (2009).
Phenotype of SGA in triploid wheat hybrids
Causal genes for the abnormal growth phenotypes observed in triploid hybrids with Ldn are widely distributed in A. tauschii (Mizuno et al., 2010). We previously identified five A. tauschii accessions that induce SGA in triploid hybrids with Ldn (Table 1, Fig. 1a,b); three from Azerbaijan, and two were originally collected from Dagestan and Iran (Mizuno et al., 2010). No phenotypic differences were observed among the resulting triploid hybrids exhibiting SGA. In this study, we additionally produced three triploid hybrids that were crossed with two tetraploid wheat accessions, T. turgidum ssp. dicoccoidesKU-8736A and T. turgidum ssp. carthlicumKU-138. The three hybrid plants, KU-8736A × A. tauschiiKU-2110, KU-8736A × A. tauschiiIG47182, and KU-138 × A.tauschiiIG47182, exhibited SGA during the winter (January) after the seeds were sown in November (Fig. 1c). All hybrid plants with the SGA phenotype extended two or three leaves without tillering until the spring, and then they died (Fig. 1a,b). Therefore, no offspring were obtained from any hybrid plants displaying SGA. It has previously been reported that high growth temperatures suppress necrotic symptoms more effectively than normal growth temperatures in hybrid necrosis of common wheat, Arabidopsis, and lettuce (Dhaliwal et al., 1986; Bomblies et al., 2007; Jeuken et al., 2009). However, the appearance of the SGA phenotype was also confirmed under high-temperature conditions (30°C) using a plant growth chamber. The SGA phenotype was not alleviated under high-temperature conditions, and the appearance of SGA was temperature-independent.
TEM observation of leaf mesophyll cells
Reactive oxygen species (ROS) generation has been observed in triploid wheat hybrids with type II and III necrosis, as well as wheat and Nicotiana plants undergoing hybrid necrosis (Mino et al., 2002; Sugie et al., 2007; Mizuno et al., 2010, 2011). To study the detailed cytological and physiological changes associated with SGA, the intracellular structures of mesophyll cells were compared among triploid hybrids formed from crosses between Ldn and PI476874 (WT), Ldn and KU-2025 (type II necrosis), and Ldn and IG47182 (SGA) using TEM. For ROS detection under TEM observation, CeCl3 was added to the fixation solution as cerium pretreatment is a highly sensitive procedure for localizing intracellular H2O2. ROS detection is based on the reaction between H2O2 and CeCl3 to produce insoluble precipitates of highly electron-dense cerium perhydroxides (Bestwick et al., 1997; Mizuno et al., 2010).
Living mesophyll cells containing large vacuoles and several chloroplasts were mainly observed in the leaves of WT hybrids and the proximal parts of leaf blades of SGA hybrid plants (Fig. 2a,c). In the living cells of SGA-exhibiting plants, black deposits corresponding to cerium signals were frequently observed within intercellular spaces, indicating the accumulation of H2O2 (Fig. 2c). Numerous fat droplets were also detected within the chloroplasts of these cells (Fig. 2d). In addition, the granum-lamella structure of living cells appeared abnormal, and the linear arrangement of grana was frequently entangled within the chloroplasts of SGA hybrid plants.
Plasmolysis, cell membrane disruption, collapse of vacuoles, and organelle degradation were typical characteristics of the dead cells in WT and SGA hybrid plants (Fig. 2b,e). Only a small number of dead cells were detected in the leaf blades of WT plants, and in the proximal parts of leaf blades of SGA plants (Fig. 2f). Although dead mesophyll cells were detected in the distal parts of leaf blades in SGA plants, the color of the leaves remained green (Fig. 2e). The degradation of intracellular structures in the dead cells of SGA plants resembled that seen in type II necrosis plants grown under normal temperature conditions (23°C) (Fig. 2b). Taken together, these observations implied that ROS generation occurred in cells before death.
Alteration of gene expression profiles in crown tissue
To comprehensively compare gene expression profiles between WT (cross between Ldn and PI476874) and SGA (cross between Ldn and IG47182) triploid hybrids, transcriptome analysis was performed using a wheat-specific 38k oligo DNA microarray (Kawaura et al., 2008). For hybridization, total RNA was extracted from crown tissues, consisting of the basal tissue of culms and tillers, shoot apices, and shoot apical meristems (SAMs), of 3-wk-old seedlings grown at normal temperature (23°C). After hybridization with the RNA samples, probes showing at least a threefold difference in signal intensity compared with the WT were defined as either up- or down-regulated genes.
Of the 37 826 probes on the wheat microarray, 3483 (9.2%) and 1914 (5.1%) probes were regarded as up- and down-regulated genes, respectively, in the crown tissues of SGA plants. Based on homology searches of the wheat EST database with probe sequences, 1986 (57.0%) and 1173 (61.3%) of the up- and down-regulated genes, respectively, were categorized into a total of 13 groups based on inferred function (Fig. 3). Of the up-regulated genes, defense-related genes were the most frequently (20.9%) encountered among the 13 groups (Fig. 3; Table S2). Genes related to signal transduction (10.7%), transport (9.7%), and metabolism (8.6%) were also abundantly expressed in the SGA hybrid plants. By contrast, genes related to photosynthesis (35.8%), stress (9.4%), defense (8.8%), and protein synthesis (7.8%) were down-regulated in the SGA plants.
Among the defense-related genes, pathogenesis-related (PR), defensin, chitinase, and peroxidase genes were highly up-regulated in the SGA hybrids (Table S2). In addition, transcripts of a number of disease-resistance genes, including Xa1 homolog and NBS-LRR-type protein genes, accumulated abundantly in the SGA plants. Thus, the up-regulation of a number of defense-related genes was characteristic of triploid hybrids with an SGA phenotype. Moreover, many WRKY-type transcription factor genes were among the up-regulated transcription factor genes in SGA hybrids (Table 2).
Table 2. List of the top 20 up-regulated transcription factor genes in the crown tissue of the severe growth abortion (SGA) triploid hybrid identified by microarray analysis
Among the 13 gene groups, photosynthesis-related genes showed the highest rate of down-regulation (Fig. 3), particularly in SGA hybrids, with the genes for ribulose-1,5-bisphosphate carboxylase (RuBisCo), chlorophyll a/b binding protein, and RuBisCo activase also included in these groups (Table S3). In addition, numerous transcription factor genes functioning in SAM and leaf primordia were down-regulated in SGA hybrids (Table 3). Particularly, the expression of gene homologs encoding KNOTTED1 (KN1)-homeobox, Myb-domain, and NAC-domain transcription factors, which are significantly associated with SAM function (Fletcher & Meyerowitz, 2000; Veit, 2004), was suppressed in the crown tissues of SGA plants. Abiotic stress-related transcription factor genes, such as wheat LIP19 bZIP and ethylene-responsive transcription factor genes, were also down-regulated in SGA crown tissues.
Table 3. List of the top 20 down-regulated transcription factor genes in the crown tissue of severe growth abortion (SGA) triploid hybrids identified by microarray analysis
To validate the microarray data, RT-PCR and quantitative RT-PCR analyses for 23 selected genes (13 up-regulated and 10 down-regulated genes) were conducted using the two triploid hybrids employed in the microarray analysis. Also, four additional triploid hybrids, consisting of two WT hybrids from crosses between Ldn and KU-2022, and between Ldn and KU-2059, and two SGA hybrids obtained from crossing Ldn with the A. tauschii accessions IG47188 and IG120866, were also included in the RT-PCR and quantitative RT-PCR analyses. Of the 23 genes from the 10 gene categories examined by RT-PCR, the gene expression levels of only two down-regulated genes were inconsistent with the microarray data (Figs S1, S2). For the other 21 genes (91.3%), differences in transcript accumulation levels between the crown tissues of the two triploid hybrids used in the microarray analysis corresponded closely to the microarray results. Moreover, comparison of the 21 gene expression levels among the six triploid hybrids indicated that the differences were consistent with the phenotypic differences between WT and SGA. Therefore, the RT-PCR analyses generally supported the microarray analysis results.
Comparison of gene expression profiles between hybrids displaying SGA and hybrid necrosis
The up-regulation of defense-related genes and down-regulation of photosynthesis-associated genes in SGA hybrids were also commonly observed in hybrid plants displaying types II and III hybrid necrosis, as reported elsewhere (Mizuno et al., 2010, 2011). To characterize the cellular responses associated with SGA in wheat triploid hybrids in greater detail, the expression levels of defense- and photosynthesis-related genes were compared between triploid hybrids with SGA and the two types of hybrid necrosis phenotypes. First, the microarray probes of defense-related genes (n =409) that were up-regulated in the crown tissues of SGA hybrids were selected and their signal intensities were compared with those obtained from previous comparative studies examining gene expression in the leaves of WT and type III necrosis hybrids and the crown tissues of WT and type II necrosis hybrids under low-temperature conditions (Mizuno et al., 2010, 2011). A significant positive correlation was detected between the gene expression profiles in the leaves of type III necrosis hybrid lines and the crown tissues of SGA hybrids (r =0.2197, P <0.001), whereas no positive correlation was observed between the crown tissues of type II necrosis and SGA plants (Table 4). Next, we selected and compared the expression of three defense-related genes, PR1, phenylalanine ammonia lyase (PAL), and WRKY3, among four triploid hybrids using quantitative RT-PCR (Fig. 4). Although type II necrosis plants and SGA hybrid plants accumulated transcripts more abundantly than WT plants, significant differences in expression levels were observed between the crown tissues of SGA hybrids and those of low temperature-treated type II necrosis. Similarly, the 267 and 316 defense-related probes that were up-regulated in types III and II necrosis hybrids, respectively, were compared with the microarray data of the SGA-exhibiting plants, and significant positive correlations were detected in both comparisons (Table 4). In the crown tissues of SGA hybrids, the up-regulation of defense-related genes displayed a similar pattern to the leaves of type III necrosis hybrids.
Table 4. Comparison of gene expression profiles among triploid hybrids with severe growth abortion (SGA), type II necrosis, and type III necrosis phenotypes
Number of probes
Target expression profile
The Pearson coefficient values were calculated based on the differences in signal intensities of two sets of the categorized probes, defense-related genes up-regulated and photosynthesis-related genes down-regulated, between the wildtype and abnormal growth hybrids.
Significant correlations: **, P <0.01; ***, P <0.001.
A markedly larger number of probes (n =420) for photosynthesis-related genes were down-regulated in the crown tissues of SGA plants than in the leaves of type III necrosis lines (n =14) and the crown tissues of type II necrosis lines under low-temperature conditions (n =51). Comparison of the signal intensities between these down-regulated photosynthesis-related probes in the tissues of the SGA, type III necrosis, and type II necrosis hybrids revealed a significant positive correlation between the expression profiles of the crown tissues in the type II necrosis and SGA hybrids (Table 4). However, the correlation between the leaves in the type III necrosis lines and the crown tissues of SGA plants was negative. Moreover, no significant correlation was observed between the 14 and 51 photosynthesis-related probes that were down-regulated in the types III and II necrosis hybrids, respectively. These results imply that the profile of down-regulated photosynthesis-related genes in SGA triploid hybrids was quite different from that in types II and III necrosis plants.
Comparison of photosynthetic activity among triploid hybrids
To examine the effects of down-regulation of photosynthesis-related genes, photosynthetic activity was compared among triploid hybrids produced from crosses of Ldn and KU-2069, Ldn and PI476874, and Ldn and KU-2022 for WT plants, and crosses of Ldn and IG47188, Ldn and IG47182, and Ldn and IG120866 for SGA plants. The photosystem II (PSII) activity of the SGA-exhibiting hybrids was significantly reduced in the first leaves of 3-wk-old seedlings compared with that of the WT hybrid plants (Fig. 5), which is consistent with the down-regulation of photosynthesis-related genes in SGA.
Suppression of meristematic activity in SGA
In the microarray analysis, we found that several transcription factor genes associated with SAM function were down-regulated in the crown tissues of SGA plants. One of the identified genes, a wheat KN1-type homeobox gene, Wknox1, has been reported to play a developmentally important role in the SAM of wheat, similar to that of maize KN1 and rice OSH1 (Takumi et al., 2000; Morimoto et al., 2005, 2009). Thus, we compared the accumulation levels of the Wknox1 transcript in the crown tissues of the four triploid hybrids by quantitative RT-PCR. Although no significant difference was observed in Wknox1 expression levels between the WT and type II necrosis plants, expression levels were significantly reduced in the crown tissues of SGA-exhibiting plants (Figs 4, S2).
Dramatic repression of cell cycle-related genes has been reported in the crown tissues of type II necrosis lines, which exhibit growth inhibition at low temperatures (Mizuno et al., 2011). Since we observed that growth inhibition was more severe in plants with an SGA phenotype than in type II necrosis plants, RT-PCR and quantitative RT-PCR analyses were performed to compare the transcript accumulation levels of seven cell cycle-related genes in the crown tissues of three WT and three SGA triploid hybrids. Compared with WT plants, transcript accumulation levels of the seven genes were significantly reduced in SGA plants (Fig. 6a). In particular, transcripts of B-type cyclin, replication protein A1, lethal(2)denticleless-like protein, and histone H4 genes were not detected in the crown tissues of the three SGA hybrids assayed by RT-PCR.
To examine mitotic cell division activity in the SAM of SGA hybrids, we first prepared longitudinal sections of shoot apices from the seedlings of WT and SGA plants, and then performed in situ mRNA hybridization analysis using a histone H4 antisense probe. The histone H4 gene is specifically expressed in the S phase of the cell cycle, and histone H4 mRNA levels have been used to study cell division activity (Gaudin et al., 2000; Yamaguchi et al., 2010). The size of shoot apices that included SAM from the SGA triploid hybrids (89.63 ± 2.32) was smaller than that of WT triploid plants (120.47 ± 5.93) (Fig. 6b–d), and the difference was statistically significant (Student’s t-test; P =0.0011). The histone H4 mRNA signals were nonuniformly distributed in the shoot apices of WT hybrid plants derived from a cross between Ldn and PI476874 (Fig. 6b). By contrast, no histone H4-expressing cells were observed in the SAM of SGA-exhibiting hybrid plants produced from a cross between Ldn and IG47182, although a similar nonuniform distribution of the histone H4 signal was found in the developing and elongating leaves of SGA plants (Fig. 6c,d). These results indicated that mitotic cell division activity is strongly suppressed in the SAM of SGA hybrid plants.
In addition, total RNA was extracted from seedlings including the crown tissues at the early developmental stage (Fig. 7a), and expression levels of the cell cycle-related genes were compared between triploid hybrids produced from crosses of Ldn and PI476874 for WT plants, and crosses of Ldn and IG47182 for SGA plants. At the early developmental stage, no SGA symptom was clearly observed, and the seedling phenotype of the SGA plants was similar to that of WT triploid plants. Compared with WT plants, transcript accumulation levels of B-type cyclin and histone H4 genes were significantly reduced in SGA plants, whereas no significant reduction of other gene expression levels was observed (Fig. 7b). These results implied that transcriptional repression of B-type cyclin and histone H4 occurs before appearance of SGA symptom.
Defective shoot meristematic activity in SGA
In the SGA-exhibiting hybrids examined in this study, plant growth was severely inhibited and then stopped before the third or fourth leaves emerged, indicating that SGA is completely lethal to seedlings (Fig. 1). To elucidate the molecular basis for the appearance of the SGA phenotype in triploid hybrids, we compared the gene expression patterns in crown tissues of the WT and SGA hybrid plants by microarray analysis. The findings revealed that a number of transcription factors functioning to maintain activity were down-regulated in the SAM of SGA plants (Table 3). Extreme repression of Wknox1 expression might significantly impair meristematic activity in the SAM. Wknox1 is an ortholog of the maize KN1 and rice OSH1 homeobox genes (Takumi et al., 2000; Morimoto et al., 2005), both of which are thought to function in maintaining SAM indeterminacy (Veit, 2004). Plants with loss-of-function mutations in KN1 orthologs fail to form or maintain SAM (Long et al., 1996; Kerstetter et al., 1997). It is therefore likely that the reduction in Wknox1 levels in the SAM of the SGA plants in this study may be associated with seedling growth inhibition.
In addition to the SAM-related transcription factor genes, cell cycle-related genes were also down-regulated in the crown tissues of SGA plants (Fig. 6). The localization pattern of histone H4 mRNA at shoot apices corresponded well with the characteristics of the SGA phenotype. At the early developmental stage, transcriptional repression of B-type cyclin and histone H4 occurs before appearance of SGA symptoms and down-regulation of other cell cycle-related genes (Fig. 7). These observations strongly suggest that severe developmental arrest in hybrids displaying the SGA phenotype may be triggered by impaired mitotic cell division and following aberrant meristematic activity in the region of the shoot apices.
The dysfunction of mitotic cell division induces both growth inhibition and necrotic cell death in Arabidopsis (Lin et al., 2007). For example, silencing of the cell cycle-regulator gene AtCDC5, a Myb-related CDC protein gene, results in severe dwarfism and necrotic symptoms (Lin et al., 2007). In AtCDC5 RNAi plants, transcript abundances of cyclin B1, histone H4, and the KN1-type homeobox gene STM were reduced, and the most striking developmental phenotype was the loss of SAM (Lin et al., 2007). A similar defect in the SAM and in mitotic cell division has been reported in an Arabidopsis tso2 rnr2 double mutant, in which extensive programmed cell death was triggered (Wang & Liu, 2006). Cell cycle progression is essential for the maintenance of the SAM, and its failure generally impairs the balance in SAM cell numbers (Lin et al., 2007). In hybrids displaying the SGA phenotype, up-regulation of defense-related genes and down-regulation of photosynthesis-related genes were observed (Fig. 3). The alteration of transcriptome in the SGA plants might be associated with the dysfunction of mitotic cell division at shoot apices.
Relationship between SGA and autoimmune response
Disease-resistance genes may play an important role in the evolution of postzygotic barriers, such as hybrid inviability, hybrid necrosis, and hybrid weakness (Bomblies & Weigel, 2007; Rieseberg & Blackman, 2010). At least one of two causal genes of Ne1-Ne2-type hybrid necrosis (type I necrosis) in common wheat is postulated to be a disease-resistance gene (Bomblies & Weigel, 2007). In type III necrosis, which phenotypically resembles type I necrosis, programmed cell death triggered by hypersensitive response-like reactions is observed and the autoimmune response results in the necrotic phenotype (Mizuno et al., 2010). Type II necrosis is also associated with a similar autoimmune response under low-temperature conditions, and the inhibition of cell division and plant growth occurs in type II necrosis-exhibiting hybrids (Mizuno et al., 2011). Two low-temperature-dependent cell death mutants, chs3 and chs4, in Arabidopsis exhibit plant growth repression; their causal genes have been identified as a disease-resistance gene and a negative regulator of defense responses against pathogens, respectively (Huang et al., 2010; Yang et al., 2010). Taken together, these previous reports imply that autoimmune responses likely play important roles in the postzygotic hybrid incompatibilities that accompany plant growth repression.
The expression profile of up-regulated defense genes in SGA hybrids was positively correlated with that in hybrids displaying type III necrosis (Table 4). ROS-producing cells were also more prevalent in SGA seedling leaves than in the leaves of WT plants (Fig. 2). However, the progressive cell death that was observed in types II and III necrosis did not clearly occur in SGA hybrids. In addition, no significant difference in the transcript accumulation levels of the mitochondrial alternative oxidase (AOX) gene Waox1a, a marker of ROS production (Sugie et al., 2007; Mizuno et al., 2010), was detected in leaves of WT and SGA plants (data not shown).
A previous study on the hypersensitive response to Pseudomonas syringae infection in Arabidopsis showed a reduction in chlorophyll fluorescence, indicating PSII damage (Alméras et al., 2003). Although seedling leaves appeared normal in SGA hybrids, the photosynthetic activities of these leaves were significantly reduced, which is typically observed in type III necrosis (Fig. 5). However, no clear positive correlation was identified in the photosynthesis-related gene expression profiles between hybrid plants displaying SGA and the hybrid necrosis phenotypes (Table 4). Therefore, the association of autoimmune response with the SGA phenotype remains unclear.
Unique feature of SGA among hybrid incompatibilities
Epistatic interactions between closely related, but divergent, genomes induce genetic incompatibilities in hybrids (Bomblies & Weigel, 2007). SGA is considered to be one of the typical phenotypes that were first observed in triploid hybrids generated from crossings between tetraploid wheat and A. tauschii (Nishikawa, 1960). Intergenomic interactions between BA and D wheat genomes result in SGA, which suggests that the relationship between the causal genes of SGA incompatibility may be explained by the DM model. Five A. tauschii accessions have been identified that genetically induce the SGA phenotype in triploid hybrids with Ldn (Mizuno et al., 2010). Distribution of the five accessions is geographically and genealogically restricted within the A. tauschii population. The limited distribution and phenotypic identity of the produced triploid hybrids suggest that the five A. tauschii accessions share a common D-genome gene which induces SGA in triploid hybrids. However, several tetraploid wheat species also carry a causal BA-genome gene associated with SGA, indicating that the causal BA-genome gene might be widely distributed in tetraploid wheat germplasm.
Although wheat hybrids with type II necrosis show a similar growth inhibition phenotype at low temperatures, this growth inhibition is more severe in SGA hybrids. The down-regulation of cell cycle-related genes clearly occurs in the crown tissues of low-temperature-treated type II necrosis lines (Mizuno et al., 2011), whereas the expression of transcription factor genes, such as Wknox1, are not affected. In addition, a small number of histone H4-expressing cells remain in the SAM of type II necrosis lines, whereas no such cells were found in the SAM of SGA plants. Therefore, the reduction of meristematic activity might be more extensive in the SAM of SGA plants.
In interploidy and interspecific crosses of Arabidopsis, F1 embryo lethality results in a postzygotic barrier (Dilkes et al., 2008; Walia et al., 2009). A similar failure of F1 seed development is observed in the interspecific hybrids of rice (Ishikawa et al., 2011). These incompatibilities during early seed development appear to be associated with maternally expressed regulatory genes (Rieseberg & Blackman, 2010). In SGA, embryo formation occurs during seed development, and F1 seed germination appears normal (Fig. 7a). Hybrid necrosis generally inhibits plant growth before the appearance of necrotic symptoms, and hybrid plants displaying necrosis are reduced in size (Bomblies et al., 2007; Alcázar et al., 2009). However, the reduction in plant growth associated with SGA is typically more severe than that in hybrid necrosis, which implies that the timing of developmental arrest in SGA-exhibiting plants likely occurs between embryo lethality and hybrid necrosis.
In conclusion, SGA in triploid wheat hybrids displays unique features, such as reduced SAM activity resulting in early developmental arrest, as well as other types of abnormal growth phenotypes, such as type II necrosis. Triploid wheat hybrids exhibiting SGA and type II necrosis phenotypes may be useful for studying the relationship between meristem function and developmental arrest in interspecific hybrids. To identify novel genes associated with postzygotic hybridization barriers, the molecular nature of the causal genes mediating SGA lethality should be elucidated in future studies.
Wheat seeds used in this study were supplied by the National BioResource Project-Wheat, Japan (http://www.nbrp.jp). We are grateful to Dr Y. Yamauchi of Kobe University for technical support for the evaluation of photosynthetic activity. This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Grant-in-Aid for Scientific Research (B) no. 21380005 and Grant-in-Aid for Challenging Exploratory Research no. 23658010), and the Sumitomo Foundation to S.T., and by a Research Fellowship from the Japan Society for the Promotion of Science for Young Scientists to N. M.