Cold stress contributes to aberrant cytokinesis during male meiosis I in a wheat thermosensitive genic male sterile line



    1. National Key Laboratory of Crop Genetic Improvement, Huazhong Agricultural University, Wuhan 430070, China
    2. Beijing Engineering and Technique Research Center of Hybrid Wheat, Beijing Academy of Agricultural and Forestry Sciences, Beijing 100097, China
    Search for more papers by this author

    1. Beijing Engineering and Technique Research Center of Hybrid Wheat, Beijing Academy of Agricultural and Forestry Sciences, Beijing 100097, China
    Search for more papers by this author
  • DI YANG,

    1. National Key Laboratory of Crop Genetic Improvement, Huazhong Agricultural University, Wuhan 430070, China
    Search for more papers by this author

    Corresponding author
    1. National Key Laboratory of Crop Genetic Improvement, Huazhong Agricultural University, Wuhan 430070, China
    2. Beijing Engineering and Technique Research Center of Hybrid Wheat, Beijing Academy of Agricultural and Forestry Sciences, Beijing 100097, China
    Search for more papers by this author

    Corresponding author
    1. National Key Laboratory of Crop Genetic Improvement, Huazhong Agricultural University, Wuhan 430070, China
    Search for more papers by this author

C. Zhao. e-mail:; Y. Zheng. Fax: +86 027 87286870; e-mail:


The male sterility of a wheat thermosensitive genic male sterile (TGMS) line is strictly controlled by temperature. When the TGMS line BS366 was exposed to 10 °C from the pollen mother cell stage to the meiosis stage, a few pollen grains were formed and devoid of starch. We report here a large-scale transcriptomic study using the Affymetrix wheat GeneChip to follow gene expression in BS366 line anthers in response to cold stress. Notably, many cytoskeletal signaling components were gradually induced in response to cold stress in BS366 line anthers. However, the cytoskeleton-associated genes that play key roles in the dynamic organization of the cytoskeleton were dramatically repressed. Histological studies revealed that the separation of dyads occurred abnormally during male meiosis I, indicating defective male meiotic cytokinesis. Fluorescence labelling and subcellular histological observations revealed that the phragmoplast was defectively formed and the cell plate was abnormally assembled during meiosis I under cold stress. Based on the transcriptomic analysis and observations of characterized histological changes, our results suggest that cold stress repressed transcription of cytoskeleton dynamic factors and subsequently caused the defective cytokinesis during meiosis I. The results may explain the male sterility caused by low temperature in wheat TGMS lines.


Recently, the utilization of thermosensitive genic male sterile (TGMS) lines is an alternative way of exploiting heterosis in wheat. The TGMS system is advantageous for its broad restoring ability, easy maintenance and multiplication (Jordaan 1996; Virmani & Ilyas-Ahmed 2001). It is also considered to be more efficient than cytoplasmic male sterility system in hybrid wheat production (Jordaan 1996; Virmani & Ilyas-Ahmed 2001). Male sterility expression in a TGMS line is heritable, but regulated by an appropriate temperature. At a certain low temperature occurring after spike differentiation, the pollen viability is dramatically altered from fertility to complete sterility (Li et al. 2006).

The anthers are the male reproductive organs of flowering plants. Within the anther, male sporogenous cells differentiate and undergo meiosis to produce microspores, which develop into pollen grains, whereas other sporophytic tissues contribute to pollen maturation, protection or dispersal (Goldberg, Beals & Sanders 1993; Scott, Spielman & Dickinson 2004; Wilson & Yang 2004). One key event during male reproductive development is the meiotic cytokinesis, which is essential for the appropriate distribution of cytoplasmic components and chromosomes to daughter cells.

Meiotic cytokinesis of plant requires a plant-specific structure known as the phragmoplast, which consists mainly of complex arrays of microtubules and mcirofilaments and expands centrifugally to the parental cell walls. Expansion of the phragmoplast is mediated by depolymerization of microtubules at the centre of the phragmoplast and polymerization of tubulin monomers at the leading edge of the phragmoplast (Jürgens 2005). Fusion of Golgi-derived vesicles occurs within and in the equatorial region of the phragmoplast, resulting in the eventual formation of new membranes and the cross walls known as cell plates (Jürgens 2005; Reichardt et al. 2007).

To date, only one mitogen-activated protein kinase (MAPK) signaling pathway with implications in cytoskeletal organization has been largely resolved. This NACK-PQR pathway plays a key role in the progression of cytokinesis in tobacco somatic cells and consists of the nucleus- and phragmoplast-localized protein kinase 1 (NPK1), the NPK1-activating kinesin-like protein 1 (NACK1) and NACK2, the Nicotiana kinase next to NPK1 MAPK kinase and the NRK1 MAPK (Soyano et al. 2003; Tanaka et al. 2004; Nakagami, Pitzschke & Hirt 2005; Beck et al. 2010). It is thought that the activated pathway ultimately targets the tobacco MAP65-1a and 1b isoforms to promote microtubules dynamics for the expansion of the phragmoplast (Sasabe et al. 2006). In the NACK-PQR pathway, NPK1 gene encodes a member of the MAPK kinase kinase family, which is preferentially expressed in proliferating and division-competent cells (Machida et al. 1998; Nakashima et al. 1998). The NPK1 is localized at the leading edge of the equatorial zone of the phragmoplast during cytokinesis (Nishihama et al. 2001). The NACK1 and NACK2 activators of the NPK1 pathway are homologous to the respective Arabidopsis genes HINKEL and STUD/TETRASPORE (TES) (Yang et al. 2003; Tanaka et al. 2004). In the Arabidopsis, the tes mutant shows male meiosis-specific cytokinesis defects only (Yang et al. 2003), while disruption of HINKEL is embryo lethal (Tanaka et al. 2004), suggesting that the two homologous kinesin-related proteins perform comparable roles in two different types of cytokinesis. There is also indirect evidence that NPK1 signaling plays a role in male meiotic cytokinesis as well (Krysan et al. 2002; Soyano et al. 2003). However, how TES/NACK2 and NPK1 signaling might affect disassembly of the radial microtubule arrays remains to be determined.

Plant reproductive development is more vulnerable than vegetative growth to many environmental stresses. In particular, cold stress causes male sterility in certain plant species (Mamun et al. 2006). Plants differ in their tolerance to chilling (0–15 °C) and freezing (<0 °C) temperature. Furthermore, studies have shown that transcriptome response to cold stress differs between organs (Kreps et al. 2002). In vegetative tissues, cold acclimation is well studied and involves a wide array of metabolic changes governed by extensive reprogramming at the gene expression level (Thomashow 1999; Fowler & Thomashow 2002; Cook et al. 2004; Hannah, Heyer & Hincha 2005; Zhu, Dong & Zhu 2007). However, little detailed work has been carried out on any of gene regulatory networks involved in chilling injury in wheat anthers.

There is considerable interest in anther-specific responses to cold stress in TGMS lines, and to identify and characterize genes that might play vital roles in anther development and are affected by cold stress. Because anthers are small and more difficult to harvest than many other plant organs, anthers are not amenable to many of the biochemical and cell biological techniques that can be used to study signaling in other plant parts. However, we overcame this problem to determine the time course of gene regulation for many physiological and developmental processes from the pollen mother cell (PMC) stage to the meiosis stage in TGMS-line anthers. Our results provide insights into a cold-induced developmental event resulting in pollen grain sterility. Genes that play key roles in dynamic organization of the cytoskeleton were dramatically repressed during meiosis under cold stress; however, transcripts involved in putative signaling cascades that lead to rearrangement of the cytoskeleton were gradually induced. Furthermore, cytological observations revealed that male meiotic cytokinesis was defective during meiosis I under cold stress, causing abnormal dyad development. The findings presented here provide insight into the possible gene regulatory mechanisms underlying male sterility in a wheat TGMS line, and improves our limited understanding of the effects of cold stress on male meiosis at the molecular level.


Plant materials

Wheat (Triticum aestivum) cv. Jing411 and TGMS line BS366 were used in this study. BS366 was selected from a natural mutant of doubled haploid lines (offspring of Jingnong8121/E8075-7). The plants were grown in soil in plastic pots embedded in the ground. Seeds were sown in excess and seedlings were thinned to a uniform population after germination. The plants were vernalized naturally in the field. After the four-leaf stage (when the tip of the fourth leaf had emerged), plants of uniform growth were selected and then randomly assigned to the experimental and control groups. The selected plants were grown in phytotrons (Koito, Tokyo, Japan) at 20 °C with a 12 h photoperiod, for the entire reproductive period, except during low-temperature treatment. Low-temperature treatment consisted of 10 °C with a 12 h photoperiod and was initiated when the flag leaf had half-emerged from the collar of the penultimate leaf (about 1.5 mm in anther length). The plants received cold stress for 5 d. Reproductive growth of individual plants and development of each spikelet in a panicle are well synchronized under controlled conditions in a growth phytotron. A diagrammatic summary of the sampling schedule is shown in Fig. 1.

Figure 1.

Diagrammatic representation of the experimental design for monitoring expression profiles in anthers and the terminal morphology of anthers and pistils in a thermosensitive genic male sterile (TGMS) wheat line exposed to low temperature. As shown in the upper panel, the TGMS lines were grown in phytotrons under the control condition of 20 °C with a 12 h photoperiod, starting at the four-leaf stage. Cold treatment, comprising 10 °C with a 12 h photoperiod, was initiated when the flag leaf had half-emerged from the collar of the penultimate leaf. (a) Under cold stress, abnormally indehiscent anthers lacked pollen grains, but pistils were morphologically normal and fertile. (b) Under cold stress, pollen grains from three squashed sister anthers were devoid of starch (stained with I2-KI). (c) Pistils and anthers of a control plant at the heading stage. (d) Pollen grains from a control plant filled with starch (stained with I2-KI).

Sample collection and morphological observation

Anthers were excised from spikelets on the primary stems with home-made dissecting needles under a dissecting microscope (Olympus SZX12, Tokyo, Japan). As shown in Fig. 1, anthers from six stages were surveyed (including the T0 stage). The anthers from the corresponding stages of control anthers were also harvested, using leaf age index and anther length as guides. To maximize the morphological synchronicity of samples at each individual stage, about 150 anthers were dissected from 50 spikelets, which were collected from 15 primary stems at each stage. At the same time, at least 40 anthers were randomly selected from the sample pool for morphological observation to control the quality of samples collected.

Microsectioning of anthers was performed to confirm the respective developmental stage to ensure accurate pooling of samples, because there was so little anther elongation between the PMC and meiosis stages (Supporting Information Fig. S1). For microsectioning, anthers were fixed in a formalin-aceto-alcohol solution and dehydrated in a graded ethanol (EtOH) series (30, 40, 50, 60, 70, 85, 95 and 100%; 1 h each). The dehydrated samples were infiltrated gradually in a Histochoice (Sigma, St Louis, MO, USA) and EtOH series (25% Histochoice: 75% EtOH; 50% Histochoice: 50% EtOH; 75% Histochoice: 25% EtOH; and 100% Histochoice). The samples were embedded in paraffin wax (Sigma) and cross sections 8 µm thick were cut with a microtome (Leica, Heidelberg, Germany). The sections were stained with 1% Safranin O solution and photographed.

Transmission electron microscopy

For transmission electron microscopy (TEM), anther pieces were pre-fixed in a solution of 2.5% glutaraldehyde adjusted to pH 7.4 with 0.1 m phosphate buffer, postfixed in 2% osmium tetroxide in the same buffer, dehydrated and then embedded in epoxy resin. Ultrathin sections (700–1000 Å) were mounted onto mesh copper grids with a diamond knife on an ultramicrotome (Leica), stained with uranyl acetate and lead citrate, and examined with a Hitachi H-7650 (Hitachi, Tokyo, Japan) TEM at 80 kV.

Fluorescent labelling of microtubules

Samples were prepared from BS366 anthers according to the method of Xu et al. (2001). Coverslips bearing stained samples were mounted on a microscope slide in anti-fade mounting medium (Beyotime, Shanghai, China) and fluorescence was examined using a Leica TCS SP2 confocal laser scanning microscope. Illustrations were captured and prepared using Leica Confocal software.

GeneChip processing and data analysis

Anther tissues at the various stages were homogenized with a Handy Pestle (Toyobo, Osaka, Japan). Total RNAs were isolated with the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. Total RNAs from duplicate biological replicates of the TGMS line were used to synthesize cRNA for hybridization with the Affymetrix GeneChip Wheat Genome Arrays (Affymetrix, Santa Clara, CA, USA). Labelling and hybridization of RNA was carried out according to standard Affymetrix protocols. In brief, 10 µg total RNAs for each replicate were converted to cDNA, from which biotinylated cRNA target was synthesized. The resulting cRNA was purified on RNeasy columns and then fragmented for hybridization to Wheat Genome GeneChips (Affymetrix). After hybridization, the arrays were washed and stained with a streptavidin–phycoerythrin conjugate on the Affymetrix Fluidics Station 400, and the GeneChips were then scanned on an Agilent GeneArray Scanner 2500A (Agilent, Palo Alto, CA, USA).

Preliminary data analysis was performed using Microarray Suite version 5 (Affymetrix), where the overall reproducibility between experiments was assessed with scatter plots of the replicates. The raw signals were imported into GeneSpring version 10 for further analysis. Per-chip normalization to the 50th percentile was applied to control for between-chip variation in expression intensities. Probe sets were considered differentially expressed with statistical significance when P-values were ≤0.05, corresponding to the false discovery rate of 5%. All further comparisons and cluster analysis were performed in GeneSpring using the default settings.

Gene annotations were obtained through batch BLASTx (E-value < e−10) searches using the Affymetrix probe sets sequences in tandem with sourcing annotations directly from the Affymetrix NetAffx Analysis Center (

Quantitative real-time PCR verification

Quantitative real-time PCR (qPCR) was performed with a MiniOpticon (Bio-Rad, Hercules, CA, USA) using the SYBER ExScript RT-PCR Kit (TaKaRa, Kyoto, Japan). Three replicates were performed for each sample along with template-free reactions as negative controls. The qPCR data were analysed using the ‘Livak’ method (Livak & Schmittgen 2001). Two internal standards (TaGAPdH and TaEFA genes) were used for each tissue stage and results were averaged over all biological replications to reduce systematic and biological variance.


Histological analysis of cold-stress injury during anther development in the TGMS line

During the PMC and meiosis stages, BS366 anthers were hypersensitive to low temperature. When BS366 lines were exposed to 10 °C for 5 d, starting at the PMC stage and continuing to the meiosis stage, complete male sterility occurred at the subsequent heading stage (Fig. 1a,b). However, anthers were fertile at 20 °C (Fig. 1c,d).

Over the course of exposure to low temperature for 5 d, the centrally located sporogenous cells developed into microsporocytes, which then underwent the meiotic divisions (Supporting Information Fig. S1). The first indication of an abnormality in cold-stressed meiocytes was the absence of a cell plate at telophase I (Fig. 2b). More severe disruption was detected in the meiotic division stage, in which defective cytokinesis with an abnormal cell plate formed as an incision observed (Fig. 2d, Supporting Information Fig. S2). In contrast, normal cytokinesis and dyad formation was more apparent in anthers in the controls (Fig. 2c, Supporting Information Fig. S2). Interestingly, before cytokinesis, we observed that the paired chromosomes were normally arranged at the equatorial plate at metaphase I under cold stress (Supporting Information Fig. S3b), and subsequently, the daughter chromosomes moved to the poles (Supporting Information Fig. S3d).

Figure 2.

Meiocytes during meiosis I showing cytokinesis under cold stress. (a) Telophase I of meiocytes in the control, with the cell plate developing (arrow). (b) Telophase I of meiocytes under cold treatment, with slight abnormality of cross wall formation (arrow). (c) Cytokinesis of meiocytes in the control, with normal cell plate formation between the two daughter cells (arrow). (d) Cytokinesis of meiocytes under cold treatment, with defective cytoplast separation (arrow). Scale bar represents 10 µm.

To elucidate structural alterations at the subcellular level induced by cold stress, we investigated anther development from the PMC stage to mature pollen stage with TEM. During tapetum development, no significant structural abnormality in the anthers from cold-treated plants was apparent, compared with those of the controls (Fig. 3a–d). Under cold stress, the tapetal cells were regular, showing dense cytoplasmic contents and numerous well-developed orbicules (Fig. 3c). The degeneration of tapetal cells proceeded gradually after meiosis and only the remnants of these cells were visible along the periphery of the anther locule at later stages of microsporogenesis (Fig. 3d). During the degeneration of tapetal cells, the microspore acquired a conspicuous wall patterning (Fig. 3c,d). Cleavage of the cytoplasm between the two daughter cells has started by the end of telophase I. During this stage, the smooth plane assembled between two daughter cells in association with cytoplasmic vesicles in normal anthers (Fig. 3e). Under cold stress, each dyad was separated from its neighbors by an undulating cell plate of uneven outline (Fig. 3f).

Figure 3.

Cytological features of anthers in the cold treatment and control observed by transmission electron microscopy. (a, b) Tapetal cells during early and late developmental stages of microspores in the control. (c, d) Tapetal cells during early and late developmental stages of microspores under cold treatment. (e) In the control, a smooth cell plate was assembled between two daughter cells (arrow) associated with numerous cytoplasmic vesicles (arrowhead) by the end of meiosis I. (f) Under cold stress, the dyads were uneven in outline, with undulating cell plates (arrow) by the end of meiosis I. CP, cell plate; Dy, dyad; En, endothecium; MI, middle layer; Msp, microspore; N, nucleus; Pg, pollen grain; Ta, tapetum; Ub, Ubisch bodies.

Thus, our light-microscopic and TEM observations suggested that male sterility was mainly due to defective cytokinesis during meiosis I, resulting in abnormal formation of dyads. Realizing that cytokinesis is the main function of the cytoskeleton during telophase I, cytoskeletal rearrangements were also studied during meiosis I. Under cold stress, the clearly visible spindles were formed as the meiocytes entered metaphase I (Fig. 4b). The spindles showed typical bipolar and highly fusiform configurations, with the chromosomes positioned at the metaphase plate (Fig. 4b). At the conclusion of anaphase I, the interzonal microtubules appeared in a tight configuration between the recently separated chromosomes (Fig. 4d). In the controls, after the two nuclei are formed at telophase I, the central microtubules are rearranged so that they form a barrel-shaped phragmoplast at the equatorial region (Fig. 4e), after which, the phragmoplast expands towards the parental cell as a hollow cylinder (Fig. 4g), with eventual formation of the cross walls known as cell plates (Fig. 4i, Supporting Information Fig. S3e). However, under cold stress, the centrifugal movement of phragmoplast was asymmetrical at telophase (Fig. 4f). During this process, the phragmoplast was continually disassembled at its center and reassembled at its periphery as aberrant C-shaped structure (Fig. 4h,j). Subsequently, the cell plate grew abnormally (Supporting Information Fig. S3f), guided by the defective phragmoplast. The microtubules were poorly defined in the defective daughter cells (Fig. 4l) and a few of these dyads were capable of entering meiosis II (Supporting Information Fig. S4).

Figure 4.

Phenotypes with defective phragmoplast in male meiosis I of BS366 anthers. Meiocytes at metaphase I in the control (a) and under cold stress (b), showing characteristic spindle fibers. Meiocytes at anaphase I in the control (c) and under cold stress (d), showing clearly visible central fibres between the two daughter nuclei. (e) Meiocytes at early telophase I in the control, with the barrel-shaped phragmoplast. (f) Meiocytes at early telophase I under cold stress, with the phragmoplast appearing in asymmetric pattern. (g) Meiocytes at mid-telophase I in the control, showing symmetric centrifugal movement of phragmoplast. (h) Meiocytes at mid-telophase I under cold stress, showing the progressive and asymmetric curvature of phragmoplast during centrifugal movement. (i) Meiocytes at late telophase I in the control, showing the formation of a ring phragmoplast, only two edges of which were visible in the focal plane. (j) Meiocytes at late telophase I under cold stress, showing the C-shaped phragmoplast. (k) Dyads at interkinesis in the control, showing radial microtubules around each nucleus and intact cytokinesis. (l) Dyads at interkinesis under cold stress, showing disorganized microtubule array and premature cytokinesis. Scale bar represents 10 µm.

Taken together, our histological results indicated that a brief episode of cold stress during the hypersensitive stage in a TGMS line mainly disrupted the dynamic organization of phragmoplast microtubules and deposition of the cell plate, causing defective cytokinesis during meiosis I.

Cold stress initiated a transcriptional cascade in TGMS-line anthers

To identify genes that play pivotal roles in anthers during cold-stress hypersensitivity, we collected anthers at six developmental stages from BS366 plants grown under cold stress to monitor the gene expression profiles. The results of cRNA hybridization with the Affymetrix GeneChip Wheat Genome Arrays for each biological replicate were highly reproducible (mean r2 value: 0.9875 ± 0.0024). Therefore, the mean expression value for each probe set at each time point was used for further analysis. The histological observations presented in Supporting Information Fig. S1 meant that interpretation of the expression data required consideration of time-series developmental changes occurring in anthers. To identify genes that were differentially expressed at a time point compared with the corresponding preceding stage, the fold-change was calculated with GeneSpring. Genes were considered significantly up- or down-regulated over the control when they showed an absolute fold-change value of 2 or greater and were assigned an adjusted P-value <0.05, corresponding to a false discovery rate of 5%.

Using these criteria, 3155 and 4883 probe sets showed altered expression in the anthers in the cold treatment and controls, respectively. The numbers of probe sets induced and repressed at each time point are shown in Fig. 5a. Relative to the T0 stage, transcripts showed sharply decreased expression in anthers at the L1.5 stage. There were 365 probe sets down-regulated, of which only about 10.0% were still expressed at later stages, which was consistent with the hypothesis that cold stress represses a range of developmental processes required for further anther development. Rather than sharply increased or decreased bursts of expression, genes underwent subtle changes in expression in the L2.0, L2.2 and L2.5 stages. Furthermore, among the up-regulated transcripts, most were stage-specific from the L1.5 stage to the L2.5 stage, which showed that the anthers responded independently to cold stress at different developmental stages. In the controls, the number of probe sets down-regulated increased steadily from 105 at the C1.5 stage to 323 at the C2.5 stage. At the same time, most of the genes showing up-regulated changes in these stages were expressed in other stages. These patterns may reflect the completion of processes involved in pre-meiotic stages, followed by down-regulation of the genes involved. The largest numbers of differentially expressed probe sets, most of which were expressed in a stage-specific manner, were found at the L3.0 and C3.0 stages. According to morphological observations, the L3.0 and C3.0 stages corresponded to the meiotic division stages (Supporting Information Fig. S1e,j). However, only 234 genes overlap between these two stages, which highlighted the destinies of anther development dramatically diverged at meiosis stage (Supporting Information Fig. S5).

Figure 5.

Cold stress initiated a differential transcriptional cascade in thermosensitive genic male sterile-line anthers, compared with the control anthers. (a) The numbers of probe sets showing transcripts induced (above the X-axis) and repressed (below the X-axis) in anthers under cold treatment (L) and in controls (C) at each developmental stage. Dark shading in the histogram indicates the number of probe sets that showed stage-specific induction or repression. Light shading in the histogram indicates the probe sets that were common to all stages. (b) The differentially expressed transcripts in anthers under cold (L) and control (C) conditions were clustered according to their expression patterns. Cluster analysis was performed in GeneSpring using the default settings.

Among the 7129 probe sets differentially expressed in anthers in the cold treatment and controls, a total of 4429 probe sets without homology annotation were removed by the annotation filtering, leaving 2691 probe sets, comprising 1937 unigenes and 299 expressed sequence tags.

In total, 840 and 1829 annotated probe sets were subjected to hierarchical clustering based on their relative transcript levels in anthers over the time course under cold-stressed and control conditions, respectively (Fig. 5b). The expression profiles of anthers in the cold treatment reinforced the hypothesis of a repressed transcriptional cascade occurring at early stages. However, in the controls, the expression profile changed subtly at early stages, followed by regulated induction or repression as anther development proceeded. In addition, the relatedness among the samples was assessed in the clustering trees. In the controls, the hierarchy of relatedness reflected the progressive developmental stages of anther in terms of transcriptome and supported the anatomical observations in Supporting Information Fig. S1. In the cold treatment, the L1.0 and L3.0 stages were the most distinctive stages. Comparison of the hierarchy of relatedness of these two treatments reinforced that cold stress strongly distorted the progressive development of anthers.

Filtering of common cold-responsive genes in TGMS-line anthers with those of the control cultivar line

The application of cold stress to vegetative tissues results in the induction of many cold-regulated (COR) genes (Hannah et al. 2005). Our previous study also identified different sets of COR genes that are involved in the cold-stress response during spike development in wheat (Yang et al. 2008). To identify genes playing key roles in anther development during exposure to cold stress in the TGMS line, it is vital to be able to identify and discount genes that are only involved in the response of COR pathways. Therefore, we chose to treat the normal cultivar Jing411 with low temperature as a control for comparison with the cold-treated TGMS line. Anthers of Jing411 treated with low temperature were harvested in parallel with the cold-treated TGMS line, which then were subjected to GeneChip analysis.

Low temperature also initiated a transcriptional cascade in Jing411 anthers (Supporting Information Fig. S6). It was noteworthy that the induction of cold-related regulons in Jing411 anthers was relatively modest compared with the induction observed in BS366 anthers. To move the genes that were also differentially expressed in Jing411 anthers, we applied two criteria for cold-specific induction or repression in BS366 anthers: (1) probe sets listed in both data sets for Jing411 and BS366 under cold treatment but not in the control BS366 data set; (2) probe sets showing down- or up-regulation at the equivalent time point in both Jing411 and BS366 under cold treatment. A total of 254 annotated probe sets were filtered from the cold-treated BS366 data set by comparison with the Jing411 control. Based on the annotation of probe sets, 24 probe sets did not meet the strict criteria used for designating cold responsiveness in BS366, so these were also filtered from the cold-treated BS366 data set. All genes filtered from the cold-treated BS366 gene list were collated as a BS366 common cold-responsive genes (BS366CCRG) data set for further comparative analysis (Supporting Information Table S1).

Another means of identifying possible COR genes in our data set was to compare our gene list with those of published data sets. A data set containing 514 genes referred to as a ‘COld Standard’ (COS) set of cold-responsive genes was downloaded from published data sets (Vogel et al. 2005). Based on homology analysis, a comparison between the BS366CCRG and COS data sets showed that about 83% of the genes included in the BS366CCRG data set also appeared in the COS data set, which indicated that filtration based on the expression profile of cold-treated Jing411 to remove COR genes of cold-treated BS366 was more efficient. These two data sets shared more genes in three functional classes – metabolism, cold-stress-induced hydrophilic proteins and transcription factors – than other classes.

Time-series transcriptome analysis of TGMS-line anthers

In our experiment, whole anthers were used for isolation of RNAs. Although the meiocytes themselves form a major proportion of the tissues in the anther, the RNA from other cell types, notably epidermal cells and the tapetum, would be included in the extracts. Therefore, our data represented gene expression from a number of different cell types with the meiocytes as the major constituent. To gain insight into the gene expression profiles of meiocytes of BS366 under cold treatment, anthers at corresponding stages of development in controls were also subjected to transcriptional analysis. The availability of these data sets gave us the opportunity to compare the expression profiles of anthers under two different conditions and identify genes that play key roles in microsporogenesis during cold treatment.

To obtain a view of gene expression changes that may be indicative of microsporogenesis processes and also general developmental shifts in both treatments, the probe sets indicated to be differentially expressed in the cold treatment and controls were subjected to self-organizing map (SOM) clustering based on their relative transcript levels in both treatments. Each gene was assigned to one of 12 expression clusters (Fig. 6). To probe the functional relatedness of the co-expressed genes, we also examined the Gene Ontology annotations for the genes in each cluster (Supporting Information Table S2).

Figure 6.

Self-organizing map (SOM) clusters of expression profiles. The probe sets differentially expressed in the cold treatment and controls were subjected to SOM clustering based on their relative transcript levels in both conditions. The number of transcripts in each cluster is specified above each SOM. The Y-axis represents normalized log2 of transcript-expression levels. The X-axis represents the anther developmental stages in the controls (C) and low temperature treatment (L).

Clusters I and II contained transcripts sharing similar expression profiles in both treatments, of which transcript levels gradually increased or decreased. Expression clusters III and IV comprised transcripts up-regulated under cold treatment; however, these transcripts showed slight changes in levels in the controls. The major difference between clusters III and IV was that the transcripts in the former cluster continued to increase in levels from the L2.0 to L3.0 stages, whereas those in cluster IV showed a significant increase only in the L3.0 stage. Clusters V and VI contained genes that showed significantly increased mRNA levels at the C3.0 stage in the controls, and differed in expression only slightly over the entire span of anther development under cold treatment. Transcripts in clusters VII and VIII showed reduced expression at the C3.0 and L3.0 stages, respectively. Clusters IX–XI contained genes whose expression gradually increased during the earlier stages of anther development and rapidly decreased at the C3.0 stage in the controls. Compared with the controls, these genes showed slight changes in transcript levels under cold treatment. A distinguishing feature of cluster XII was that transcripts increased dramatically between the C1.5 and C2.5 stages and then decreased sharply at later stages, whereas there was a graduated increase in the level of these transcripts at later stages under cold treatment.

The results of SOM clustering indicated that most of the transcripts showed differential expression patterns in anther development between the cold treatment and controls, except for those of clusters I and II, which indicated that anther development proceeded abnormally under cold treatment. To address this consideration, we investigated the possible functions of genes in the full set of clusters. Cluster I contained a number of transcripts with annotations to genes known to have roles in tapetum development, such as Undeveloped Tapetum1 (Udt1, TaAffx.72045.1.A1_at) (Jung et al. 2005) and Triticum aestivum sporophytically produced structural protein (TaRAFTIN) (Ta.4176.1.S1_x_at) (Wang et al. 2003). Similarly, it was noteworthy that cluster II contained genes involved in a variety of aspects of meiosis control, including cyclin-dependent kinases1 (Cdk1, Ta.7655.1.S1_at), cyclin B (Ta.14556.1.S1_x_at, Ta14556.2.A1_at), S-phase kinase-associated protein-1 (Ta.23791.1.S1_x_at), Cdh1/Hct1 (Ta.27562.1.S1_at), structural maintenance of chromosome protein 4 (Ta.6708.1.A1_at) and disrupted meiotic cDNA1 (Ta.30833.1.S1_at) (Yang et al. 1999; Cobbe & Heck 2004; Neale & Keeney 2006; Pesin & Orr-Weaver 2008). Therefore, genes in cluster II may preferentially regulate the meiotic process in microsporocytes. The genes in cluster III were COR genes and responded to cold treatment, of which 83% were listed in the BS366CCRG data set mentioned above. Interestingly, a number of known cytoskeleton-associated genes were identified in clusters IV, VI, VIII and X, such as profilin, formin, myosins and kinesin, which will be discussed in more detail in the following section.

Expression of several cytoskeleton-associated genes was affected by cold stress in TGMS-line anthers

Analysis of the current data sets using SOM clustering indicated that cold stress induced or repressed the levels of transcripts putatively involved in changing the organization and distribution of the cytoskeleton, including nine genes belonging to a variety of cytoskeleton-associated factors, and six genes involved in signal-transduction cascades that affect cytoskeletal reorganization (Table 1).

Table 1.  The fold-change level of the cytoskeleton-associated genes in anthers under cold treatment compared with its counterpart stages of control treatment
Probe setL1.5 versus C1.5L2.0 versus C2.0L2.2 versus C2.2L2.5 versus C2.5L3.0 versus C3.0Annotation
  1. CDPK, calmodulin-like domain protein kinase; ROP, Rho of plants; NPK1, nucleus- and phragmoplast-localized protein kinase 1; ADF, actin-depolymerizing factor.

Ta.6292.2.S1_a_at1.176Up1.353Up1.023Up1.339Up2.855UpCDPK 13
Ta.11095.1.A1_at1.414Up1.212Up1.597Up1.26Up2.559UpROP GNE
TaAffx.37308.1.A1_at1.482Up2.968Up1.134Up1.096Down1.057DownSCAR-like protein
Ta.14482.1.S1_at1.444Up1.094Up1.151Down1.12Down4.433DownLIM domain protein
Ta.1396.1.S1_at1.122Down1.558Down1.883Up1.273Up5.383UpROP-like protein
Ta.19020.1.S1_at2.165Down1.152Down1.187Down1.909Up3.375UpNPK1-like protein
Ta.853.2.S1_at1.031Down1.488Down1.102Down1.176Down2.005DownMyosin XI

Profilin binds with high affinity to monomeric actin and functions at the center of dynamic actin assembly. In the presence of uncapped barbed ends of microfilaments, the profilin–actin complex shuttles actin subunits onto pre-existing filaments (Staiger & Blanchoin 2006). Some of the proline-rich protein partners, such as formins, can use profilin–actin to drive actin filament assembly directly (Deeks, Hussey & Davies 2002). Interestingly, in cold-treated anthers, the profilin homology transcript (Ta.15371.1.S1_at) was repressed 27-fold at the L3.0 stage, compared with the C3.0 stage (Table 1). Similarly, the ADF (actin-depolymerizing factor) gene homologous transcript (Ta.24728.1.S1_at) was dramatically repressed sevenfold at the L1.5 stage and moderately repressed in subsequent stages under cold treatment (Table 1). ADF acts synergistically with profilin to affect actin dynamics (Staiger & Blanchoin 2006). Furthermore, phosphorylation is a principal regulator of ADF activity, and a calmodulin-like domain protein kinase (CDPK) is known to phosphorylate plant ADF (Smertenko et al. 1998; Allwood, Smertenko & Hussey 2001). Interestingly, a transcript of CDPK gene homology (Ta.6292.2.S1_a_at) was induced following cold treatment and dramatically changed 2.9-fold at the L3.0 stage (Table 1). Fimbrin and villin act as microfilament-bundling or cross-linking proteins in plants (Huang et al. 2005). The transcript of fimbrin homology (Ta.1825.1.A1_at) was found to only slightly change in level with cold treatment; however, villin homologous transcript (TaAffx.31119.1.S1_at) was induced 1.9-fold at the L3.0 stage (Table 1). The induction of villin might prevent the ADF from binding to and severing the microfilament, which suggests that the increased villin-bundled microfilament arrays might be less dynamic than individual microfilaments. Therefore, based on the expression profiles of profilin, ADF gene and villin homologous transcripts, we inferred that actin dynamics were affected in TGMS-line anthers subjected to cold stress. Besides actin dynamic behavior, actin nucleation is required for microfilament elongation and construction of higher-order microfilament-based structures. To date, two actin nucleators have been identified in plants: the Arp2/3 complex and formins. The Arp2/3 complex initiates a new F-actin branch at an angle of 70° relative to the parent microfilament (McKinney, Kandasamy & Meagher 2002). Instead of the dendritic microfilament, the microfilament arrays are apparently organized predominantly in bundles or cables in most plant cells (Staiger 2000). The formins are good candidates to initiate and generate such structures using a plant profilin–actin complex (Cvrckováet al. 2004; Wasteneys & Yang 2004). In our data set, the transcript of the Arp2/3 complex subunit, Arp3 (Ta.24619.1.S1_at), was slightly down-regulated; however, two formin homologous transcripts (Ta.9719.2.A1_at, TaAffx.5150.1.S1_at) were repressed following cold treatment and transcript levels were repressed 2.3- and 2.5-fold at the L2.5 stage, respectively (Table 1). Tobacco Widely expressed LIM protein 1 (WLIM1), containing two Lin11-Isl1-Mec3 (LIM) domains, was identified as a novel actin-binding protein that increases actin cytoskeleton stability by promoting microfilament bundling (Thomas et al. 2006, 2007). Similarly, WLIM1 homologous transcripts (Ta.14482.1.S1_at) were dramatically repressed 4.4-fold at the L3.0 stage (Table 1).

The transcript (Ta.7152.2.S1_at) encodes a C-terminal type KIF 2 (KIFC2)2/KIFC3-like C-terminal kinesin, which has been implicated in motility of the Golgi apparatus and is required for phragmoplast growth and cytokinesis (Hirokawa 1998; Strompen et al. 2002; Tanaka et al. 2004; Lee, Li & Liu 2007), was repressed 2.6-fold at the L3.0 stage (Table 1). Furthermore, the transcript homologous with the myosin XI gene (Ta.853.2.S1_at) was repressed 2.0-fold at the L3.0 stage following cold treatment (Table 1). Myosins are a superfamily of molecular motor proteins responsible for microfilament-based motility in eukaryotes (Jiang & Ramachandran 2004). The class XI myosins were considered likely to be involved in organellar movement, particularly of mitochondria (Van, Köhler & Verbelen 2002), endoplasmic reticulum (Yokota et al. 2009), Golgi apparatus (Nebenführ et al. 1999) and vacuoles (Higaki et al. 2006).

Rho of plants (ROP) proteins are influential initiators of cytoskeletal rearrangement in plant cells (Deeks et al. 2002; Drøbak, Franklin-Tong & Staiger 2004). ROP proteins turn on signaling pathways by switching from a guanosine diphosphate (GDP)-bound inactive form to a guanosine 5′-triphosphate (GTP)-bound active conformation (Bourne, Sanders & McCormick 1991). Activation depends on guanine nucleotide exchange factors (GEFs) that catalyse the otherwise slow GDP dissociation for subsequent GTP binding (Cherfils & Chardin 1999). In the present study, transcripts of ROP gene (Ta.1396.1.S1_at) and GEF gene (Ta.11095.1.A1_at) homologies were constantly induced in anthers under cold treatment and dramatically increased in level 5.4- and 2.6-fold at the L3.0 stage, respectively (Table 1), which likely indicated that an extracellular cold-induced signal was transferred into anther cells. The suppressor of cAMP receptor (SCAR) complex is an effector of ROP cytoskeletal reorganization in plants (Li et al. 2004; Saedler et al. 2004; Zhang et al. 2005). One member of the putative plant complex, specifically Rac1-associated protein 1, binds the active form of ROP, whereas plant SCAR, another component, can activate the Arp2/3 complex in vitro and binds G-actin (Frank et al. 2004; Le et al. 2006). In cold treatment, transcript level of a SCAR-like gene (TaAffx.37308.1.A1_at) was induced 3.0-fold at the L2.0 stage (Table 1); however, Arp3 was continually repressed following cold treatment as mentioned above. The ROP-interacting Cdc42/Rac-interactive binding (RIC) motif protein is another effector of ROP cytoskeletal reorganization (Wu et al. 2001; Gu et al. 2005; Hussey, Ketelaar & Deeks 2006; Fu et al. 2009). Recent research showed that ROP2 and ROP4 in Arabidopsis promote the protrusion by activating the localized actin accumulation and inhibiting microtubule organization by inactivating the RIC proteins (Fu, Li & Yang 2002; Fu et al. 2005). Interestingly, a transcript of RIC gene homology (Ta.30457.1.A1_at) was dramatically repressed by 4.4-fold at the L3.0 stage by cold treatment (Table 1).

In addition, expansion of the phragmoplast and cell plate during cytokinesis requires the activation of a MAPK signaling pathway mediated by the formation and targeting of NACK-NPK complex to the phragmoplast (Nishihama et al. 2001; Jin et al. 2002; Krysan et al. 2002). In the present study, the transcript homologous of the NPK1 gene (Ta.19020.1.S1_at) was dramatically increased 1.9- and 3.4-fold at the L2.5 and L3.0 stages, respectively (Table 1).

The possible pathway of cytoskeleton dynamic organization and signaling, along with the expression profiles of genes that showed differential expression between the cold treatment and controls, are presented in Supporting Information Fig. S7. The genes listed in Table 1 were selected to validate representative GeneChip results using qPCR. The results confirmed a high degree of reproducibility between these two platforms (Supporting Information Table S3).


The male sterility of wheat TGMS lines is strictly controlled by temperature. When a TGMS line is exposed to an appropriate low temperature, starting at the PMC stage and continuing to the meiosis stage, TGMS lines exhibit indehiscent anthers and lack pollen grains, while other organs, including pistils, and seedlings develop normally (Fig. 1). This suggests that microsporogenesis is much more sensitive to cold stress compared with female reproductive development and vegetative growth.

Our objective is to elucidate how cold stress influences the microsporogenesis process. As a first step towards achieving this aim, the histological features were examined in TGMS-line anthers. During cold stress, abnormal separation of dyads occurred during male meiosis I, owing to failure of male meiotic cytokinesis (Fig. 2d). The fluorescence labelling and TEM results further revealed that the phragmoplast structure was defectively formed (Fig. 4f,h,j) and the cell plate was abnormally assembled during meiosis I under cold stress (Fig. 3f, Supporting Information Fig. S3f). To gain further insight into the developmental transition throughout microsporogenesis underlying the histological changes, genomic transcriptional profiling of TMGS-line anthers over a developmental time series was monitored under cold stress and in controls. A total of 840 annotated probe sets showed significant changes in transcript levels under cold treatment, but 278 of these common cold-responsive genes were eliminated by comparisons with the cold-treated Jing411 controls. Comparing the transcriptomes of cold-stressed lines and the controls, signaling systems that orchestrated cytoskeleton activity were moderately induced in response to cold stress in anthers of the TGMS line; however, genes involved in dynamic organization of the cytoskeleton were dramatically repressed.

As discussed in more detail in the following sections, the transcript profiling generated in this study, combined with the histological observations, was designed to help decipher the male sterility of the TGMS line at a molecular level.

Tapetal development was unaffected by cold stress

It is evident that meiosis and subsequent development of the male gametes are regulated in part not only by the developing male gametophytes, but also by the surrounding sporophytic tissues, especially the tapetal tissue. The tapetum plays a critical role in the nutrition of microspores and in pollen wall development.

Abiotic stresses during the PMC stage lead to aborted microsporogenesis and male sterility, which cause early degeneration or hypertrophy of the tapetum (Oshino et al. 2007; Gothandam, Kim & Chung 2009). However, adaptations of anthers to environmental stresses are quite different in plants. Our histological observations indicated that there was no appreciable structural difference in tapetum development in the control or cold-treated anthers. Under cold stress, the tapetal layer degenerated gradually after meiosis, pro-orbicules (the precursors of orbicules) were observed in the tapetum during the late meiosis stage and the gametophytic exine was well developed (Fig. 3c,d). These observations were reinforced by the transcriptional profiling analysis. Cluster I (Fig. 6), in which the genes shared the same expression profiles under control and cold-stressed conditions, contained the genes important for tapetal development, such as Udt1 and TaRAFTIN. In the Udt1 mutant, the anther walls and meiocytes were normally developed during the early pre-meiosis stage, but the tapetum failed to differentiate and became vacuolated during the meiotic stage (Jung et al. 2005). This report indicated that Udt1, a basic helix-loop-helix transcription factor, is required for differentiation of secondary parietal cells into mature tapetal cells. The RAFTIN is essential specifically for the maturation phase of pollen development. The locales of RAFTIN suggest that RAFTIN is synthesized in the tapetum, packaged in Ubisch bodies, and secreted at appropriate developmental stages to the microspores. In RAFTIN-silenced lines, tapetal degeneration was arrested apparently at the vacuolated microspore stage, and the collapsed microspores were manifested as flat pollen grains (Wang et al. 2003).

A large number of Arabidopsis mutants are either male sterile or have reduced male fertility, and at least eight of the corresponding genes determining tapetal development have been identified (Ariizumi et al. 2004; Wilson & Yang 2004; Guan et al. 2008). However, only the male sterility 2 gene, belonging to the fatty acyl-Coenzyme A reductase family, showed moderate repression by cold stress in our analysis relative to control anthers.

Taken together, several lines of evidence indicated that the functioning of the tapetum layer was largely unaffected by cold stress in TGMS-line anthers.

Cold stress contributed to aberrant cytokinesis during dyad development

It is important that initiation of cytokinesis and cell plate formation does not occur until the chromosomes have begun to segregate at the onset of anaphase, a process that requires the temporal coordination of Cdk1 inactivation, chromosome segregation and cytokinesis (Niiya et al. 2005). This temporal control of cytokinesis is achieved by a combination of overlapping mechanisms. Components of the cytokinesis machinery are sequestered in the nucleus of late S-phase cells prior to breakdown of the nuclear envelope. Then, as cells enter meiosis, Cdk1 phosphorylation inhibits many of the key players, preventing them from forming complexes that promote cytokinesis. At the onset of anaphase, these inhibitory phosphorylations are removed, which promotes the assembly of protein complexes required for cytokinesis. In all organisms, the onset of anaphase correlates with a decline in Cdk1/cyclin B1 activity, and persistent Cdk1/cyclin B1 activity has been shown to interfere with central spindle formation and cytokinesis (Wheatley et al. 1997; Barr & Gruneberg 2007). In our data sets, the transcript levels of Cdk1 and cyclin B1 genes were gradually decreased in TGMS-line anthers under cold stress, which shared similar expression profiles with those of control anthers (cluster II, Fig. 6). These results indicated that the male meiocytes of the TGMS line under cold stress were capable of entering anaphase I and forming central spindle fibers after the separation of chromosomes, which was consistent with histological observations (Fig. 4b,d).

Most monocotyledonous plants, such as T. aestivum, undergo successive cytokinesis after anaphase I, which includes formation of phragmoplast microtubules arrays, expansion of the phragmoplast towards the parental cell walls and fusion of Golgi-derived vesicles at the equatorial region, with eventual deposition of the cell plates (Tanaka et al. 2004).

According to our observations, under cold stress, male meiotic cytokinesis was defective during meiosis I (Fig. 2d), as a result of aberrant C-shaped phragmoplast (Fig. 4j). These observations indicated that the characteristic centrifugal movement of phragmoplast during telophase was severely affected by cold stress.

The molecular mechanism of phragmoplast formation in male meiocytes has not been described in detail for monocotyledons. In mitotic cytokinesis, radial actin fibers connecting the phragmoplast midzone with the division site in the cortical cytoplasm zone are considered to guide centrifugal movement of the central fibres (Baskin & Cande 1990). Because the structure and function of the phragmoplast during successive cytokinesis in meiosis in monocotyledons are analogous to those in mitosis, it is possible that radial actin fibers also play a role in modeling the progressive curvature of central fibres in meiosis. In our data sets, the actin- and microfilament-associated genes were repressed at the L2.5 and L3.0 stages. The L2.5 and L3.0 stages represented the leptotene–zygotene transition stage of meiotic prophase I and the meiotic division stage, respectively (Supporting Information Fig. S1d,e). During the L2.5 stage, the transcript levels of two formin homologs were dramatically repressed (2.3- and 2.5-fold) in anthers under cold stress, compared with the corresponding stage in control anthers (Table 1). Similarly, transcription of the profilin homolog was severely repressed (27-fold) at the L3.0 stage (Table 1). Formin combines with profilin as an actin nucleation factor and plays a key role in the modulation of microfilament bundling as well as dynamicity (Wallar & Alberts 2003; Goode & Eck 2007). As mentioned above, the radial actin fibres, connecting with the phragmoplast midzone, play a role in guiding the centrifugal movement of the central spindle fibers. However, under cold stress, the unscheduled repressed activities of formin and profilin may affect the dynamicity of microfilament bundling, and thus cause abnormally curved spindles during telophase. Furthermore, transcription of the NPK1 homolog was dramatically increased 1.9- and 3.4-fold at the L2.5 and L3.0 stages, respectively, compared to the corresponding stage of control anthers (Table 1). The NPK1 protein is localized at the leading edge of the equatorial zone of the phragmoplast and regulates lateral expansion of the phragmoplast during cytokinesis (Nishihama et al. 2001; Krysan et al. 2002; Soyano et al. 2003). Overexpression of a kinase-defective mutant form of NPK1 in tobacco cells results in inhibition of lateral expansion of the phragmoplast and formation of multinucleated cells with incomplete cell walls (Nishihama et al. 2001). There is also indirect evidence that NPK1 signaling plays a role in male meiotic cytokinesis as well (Krysan et al. 2002; Soyano et al. 2003). Thus, it is probable that the abnormally expressed cytoskeleton-related genes were major contributors to the aberrant phragmoplast during telophase in male meiocytes under cold stress.

As the phragmoplast expands centrifugally, the cell plate and its limiting membranes are formed by fusion of membrane-bound vesicles thought to originate from the Golgi apparatus. Although microtubules are generally thought to be responsible for vesicular transport, a role for microfilaments cannot be ruled out. Decoration of phragmoplast microfilaments with heavy meromyosin reveals that the opposing arrays of microfilaments are in proper orientation for transport of vesicles to the developing cell plate. In addition to the abnormal expression of genes involved in microfilament bundling mentioned above, the myosin homolog transcript was also repressed under cold stress (Table 1). Furthermore, the transcript encoding a KIFC2/KIFC3-like C-terminal kinesin, which have been implicated in the motility of the Golgi apparatus and are required for phragmoplast growth and cytokinesis, was repressed 2.6-fold at the L3.0 stage relative to the corresponding stage of control anthers (Table 1). These results indicated that vesicle fusion generating a membranous network between the two dyads during meiosis I was affected by cold stress, which was reinforced by the subcellular observations. Under cold stress, perturbed cell plates of dyads, which failed to flatten and stiffen, were exceptionally undulating (Fig. 3f). In addition, few vesicles associated with the cell plate were observed in subcellular examinations (Fig. 3f), in contrast with those in dyads of control anthers (Fig. 3e).

Due to the aberrant cytokinesis during dyad development, the microtubules were disorganized and did not define the radial system around the nucleus (Fig. 4l). Eventually, the microsporogenesis was arrested at meiosis I stage.

Data from histological observations and differential gene expression analysis, as well as comparison with known gene functions, have allowed formulation of a hypothesis concerning male sterility of a wheat TGMS line under cold stress, and to identify several functional candidate genes as potential regulators of male sterility. Clearly, the hypothesis requires functional verification using molecular genetic techniques in the future.


The authors thank the two anonymous referees for their constructive comments that have improved the presentation of this manuscript. This study was supported by the National Science Foundation (No. 30840050), National ‘863’ Program (No. 2011AA10A106), the Program of Beijing Basic Research and Innovation Platform for Agricultural Breeding (No. D08070500690801) and Beijing Science Foundation (No. 5091001).