A wheat K-type CMS line (Yumai 3, KA), a maintainer line (Yumai 3, KB), a restorer (Yumai 2) line, and F₁ lines were grown in the experimental field of the Institute of Genetics and Developmental Biology, Chinese Academy of Sciences (Beijing, China) until the heading stage. Anthers at the mononucleate, dinucleate, and trinucleate stages were sampled, quickly frozen in liquid nitrogen, and stored at -80°C until RNA isolation.

Wheat variety Chinese Spring and 21 Chinese Spring nullisomic-tetrasomic lines were grown in a greenhouse at 22°C under a 16-h light/8-h dark regime. Young leaves were collected for DNA extraction. Roots, stems, flag leaves, and young spikes at the heading stage were sampled for RNA isolation.

Arabidopsis wild-type Columbia (Col-0) and transgenic pEarleyGate 101 plants (pEarleyGate 101-TaAG-A and pEarleyGate 101-TaAG-B) were grown in a greenhouse at 22°C under a 16-h light/8-h dark regime. Leaves were collected for RNA isolation.

Normal wheat and wheat inoculated with RNA_γb:TaAG-A or RNA_γb:TaAG-B were grown in a greenhouse at 25/15°C under a 16-h light/8-h dark regime. Spikes were collected for RNA isolation.

Genomic DNA was extracted using the cetyltrimethylammonium bromide (CTAB) method (Saghai-Maroof et al., 1984). Total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, United States) and treated with RNase-free DNaseI (Promega, Madison, WI, United States) as described by Huang et al. (2012). First strand cDNA was synthesized using the FastQuant RT Kit (with gDNase) (TIANGEN, Beijing) following the manufacturer’s instructions.

Ta.4147 was used to conduct a BLAST search of the Triticum urartu genome database¹. Then two pairs of gene specific primers, TaMADS1.CF/CR and TaMADS2.CF/CR (Supplementary Table S1), were designed and used to clone MADS-box genes from the cDNA of anthers from the male sterile line at the mononucleate stage. The following PCR reaction conditions were used: 5 min at 94°C; followed by 35 cycles of 30 s at 94°C, 30 s at 58°C, and 1 min at 72°C; with a final extension at 72°C for 10 min using LA-Taq DNA polymerase (Takara, Biotechnology Co., Ltd., Dalian, China). The PCR products were individually cloned into pGEM-T easy vectors (Promega, Madison, WI, United States) and introduced into Escherichia coli. Ten positive independent clones were commercially sequenced. The GenBank/EMBL accession numbers are as follows: KX354939–KX354940. The sequences were analyzed using DNAMAN software and the Conserved Domain Search in NCBI². The phylogenetic trees were drawn using Clustal W and MEGA software.

Nulli-tetrasomic lines of Chinese Spring were employed to determine the chromosomal locations of the TaMADS1 and TaMADS2 genes with the gene-specific primers TaMADS1.LEF/LER and TaMADS2.LEF/LER (Supplementary Table S1). The PCR products were resolved by electrophoresis on a 1% agarose gel containing ethidium bromide and visualized under a UV spectrometer.

Quantitative real-time PCR (qRT-PCR) was performed to study relative gene expression levels using the LightCycler 480 system (Roche, Indianapolis, IN, United States) with SYBR Green I Master (Roche) according to the manufacturer’s protocol. Two pairs of gene-specific primers, TaMADS1.LEF/LER and TaMADS2.LEF/LER (Supplementary Table S1), were used for gene expression analysis with the Ta4045 gene, which encodes ubiquinol-cytochrome C reductase iron-sulfur subunit as described by Paolacci et al. (2009), as a control. The expression profiles of TaAG-A and TaAG-B in various wheat organs (roots, stems, flag leaves, and young spikes) and at different stages (mononucleate, dinucleate, and trinucleate) were investigated. The qRT-PCR experiment was performed independently with three biological replicates and three technical replicates according to the method of Li et al. (2013).

The entire coding sequences of the TaAG-A and TaAG-B genes were individually cloned into the pEarleyGate101 vector using the Gateway-compatible vector cloning system (Invitrogen, Carlsbad, CA, United States) to produce the plasmid constructs pEarleyGate 101- TaAG-A and pEarleyGate 101- TaAG-B. The constructs were used to transform Arabidopsis wild-type Col-0 by the Agrobacterium- mediated floral-dip method (Clough and Bent, 1998). Transgenic lines were selected on MS (Murashige and Skoog, 1962) plates with phosphinothricin (5 μg/L), and more than 20 transgenic lines were obtained. Transgenic plants were selected for phenotypic analysis and semi-quantitative RT-PCR with Actin as a control. The primers used for plasmid construction (TaAG-AB.101F/101R), transformant identification (101.F/R), and semi-quantitative RT-PCR (TaAG-AB.SF/SR, Actin.F/R) are listed in Supplementary Table S1.

The NheI restriction site was introduced into TaAG-A and TaAG-B using the primers TaAG-A-NheI-F/R and TaAG-B-NheI-F/R (Supplementary Table S1). The amplified fragments were used to replace the GFP coding sequence in RNA_γb:GFP, resulting in RNA_γb:TaAG-A and RNA_γb:TaAG-B. The preparation of BSMV transcripts for wheat inoculation was consistent with the method of Zhou et al. (2007). The fourth leaf and spikes of inoculated plants were observed at 12, 16, and 20 days after inoculation. The relative transcript levels of the TaAG-A and TaAG-B genes in the spikes were assessed by semi-quantitative RT-PCR, which were performed independently with three biological replicates and three technical replicates.

Our previous microarray results showed that the expression level of the MIKC-type MADS-box transcription factor Ta.4147 in the CMS line was higher than in the maintainer line. Therefore, the Ta.4147 sequence was used to conduct a BLAST search of the T. urartu genome database and we obtained a very similar sequence, TRIUR3_34584 (GenBank accession number KD145636). We designed gene-specific primers (Supplementary Table S1) and performed RT-PCR analysis with the purified RNA extracted from uninucleate pollen of the wheat K-type CMS line. The PCR products were separated on a 1% agarose gel and purified with the Agarose Gel DNA Purification kit (TIAGEN, Beijing) according to the manufacturer’s instructions. The purified fragment was then cloned into the pGEM T easy vector (Promega) and transformed into E. coli TOP10. Positive clones were selected for sequencing. The results of sequence analysis (Supplementary Table S2) revealed two types of sequences derived from positive clones. One sequence had an open reading frame (ORF) of 819 bp encoding a peptide of 272 amino acid residues with a molecular weight of 30.8 kDa and a pI of 9.072. The other type had an ORF of 831 bp encoding a peptide of 276 amino acid residues with a molecular weight of 31.4 kDa and a pI of 8.794. The predicted proteins (Figure 1) have conserved MADS-MEF2-like and K-box domains, which are typical characteristics of MIKC-type MADS-box proteins, there are ten amino acids different between these two MADS-box proteins

To assign the two MADS-box genes to the wheat genome, we used Chinese Spring nulli-tetrasomic lines for chromosomal localization. Specific primers for TaMADS1 and TaMADS2 were designed (Supplementary Table S1) and PCR was performed using DNA from the nulli-tetrasomic lines as templates. The results (Supplementary Figure S1) showed that N3A/T3B had no TaMADS1 band and the N3B/T3D lane had no TaMADS2 band, indicating that TaMADS1 and TaMADS2 were localized on wheat chromosomes 3A and 3B, respectively, thus these were designated as TaAG-A and TaAG-B.

We performed phylogenetic analysis using amino acid sequences derived from rice, Arabidopsis, and wheat MADS-box genes. The results (Figure 2A) showed that TaAG-A and TaAG-B belong to the AG subfamily C group of MADS-box family proteins and have very high identity to MIKC-type MADS-box transcription factors WM29B (CAM59077) and TaAGL39 (ABF57939) from wheat and 93% identity with OsMADS3 (ACY26070) from rice.

Further phylogenetic analysis of AG subfamily proteins (Figure 2B) from wheat, barley, rice, maize, Brachypodium distachyon, T. urartu, and Aegilops tauschii showed that TaAG-A and TaAG-B are closely related to AesWAG-2 (AEM24901) from Aegilops tauschii and HvAG1 (AAL93196) from barley.

To assess the expression patterns of TaAG-A and TaAG-B in wheat, we performed qRT-PCR analysis using RNA extracted from roots, stems, leaves, and young spikes of Chinese Spring wheat. Anthers were collected from the wheat K-type CMS line at the uni-, di-, and tri-nucleate stages and the maintainer line. The results (Figure 3A) demonstrate that TaAG-A and TaAG-B had similar expression patterns in different wheat tissues, but their expression levels differed in different tissues. TaAG-A had very high expression levels in young spikes and leaves and very low expression levels in roots, while TaAG-B was most abundant in young spikes with very low expression in roots, stems, and leaves.

The expression levels of TaAG-A and TaAG-B also showed similar trends at different developmental stages in anthers of the same wheat variety (Figure 3B). Both genes could be detected at all developmental stages in anthers, but TaAG-B transcripts were more abundant than those of TaAG-A. From the di-nucleate to tri-nucleate stages, the expression levels of TaAG-B were upregulated in the CMS line, but downregulated in the maintainer line. Additionally, the expression levels of TaAG-A and TaAG-B in the CMS line at the trinucleate stage were higher than in the maintainer line. These results demonstrate that TaAG-A and TaAG-B have different expression patterns between the CMS line and maintainer line at different anther developmental stages, especially at the trinucleate stage. Higher transcription of TaAG-A and TaAG-B may be closely related to male sterility in the CMS line.

To further investigate the functions of the TaAG-A and TaAG-B genes, we constructed overexpression plasmids (pEarleyGate 101-TaAG-A and pEarleyGate 101-TaAG-B) using the Gateway-compatible vector cloning system. Plasmid constructs were individually transformed into Arabidopsis wild-type Col-0. More than 20 transformants were obtained and T3 homozygous plants were used for phenotypic analysis (Figure 4A). We compared the vegetative growth of Col-0 and the pEarleyGate 101 transformants. Both the pEarleyGate 101-TaAG-A and pEarleyGate 101-TaAG-B transgenic plants were emaciated, dwarfed, and abnormal with inwardly curled young leaves, tiny yellow leaves, and advanced reproductive growth. At the ripening stage, pEarleyGate 101-TaAG-A and pEarleyGate 101-TaAG-B plants displayed abnormal growth and development, and exhibited delayed flowering and podding.

We observed the buds of transgenic Arabidopsis under a stereomicroscope (Figure 4B) in comparison with Col-0 buds. The pEarleyGate 101-TaAG-B plants had short and abnormal sepals, petals, and stamens; expanded pistils; and no selfing fructification.

We performed semi-quantitative RT-PCR analysis of flowers from transgenic plants compared with normal plants using Actin expression as a control (Figure 4C). TaAG-A and TaAG-B transcript levels were higher in plants with abnormal phenotypes, indicating that TaAG-A and TaAG-B expression had a great influence on pistil and stamen development of transgenic Arabidopsis. In addition, TaAG-B induced abnormal development more readily than TaAG-A in transgenic Arabidopsis.

To obtain more direct evidence of the functions of TaAG-A and TaAG-B, we performed virus-induced gene silencing (VIGS) developed with the barley stripe mosaic virus (BSMV) to decrease their transcript levels. Uninfected wheat plants, wheat plants infected by RNA_γb, RNA_γb:TaPDS, and RNA_γb:GFP served as controls in the inoculation assays. The fourth leaf (Figure 5A) and spikes (Figure 5B) of inoculated plants were observed at 12, 16, and 20 days after inoculation. The leaves of the wheat plants infected by RNAγb had striped spots, while wheat plants inoculated with BSMV:TaPDS had white leaves, and the leaves of BSMV:GFP-inoculated plants produced yellow, diseased spots, indicating that the BSMV experiments were successful. In plants inoculated with BSMV:TaAG-A and BSMV:TaAG-B, leaf surfaces had striped diseased spots, which were more obvious at 16 and 20 days after inoculation.

The VIGS assays also provided a visual indication of the occurrence of TaAG-A and TaAG-B gene silencing in spikes after virus inoculation. The spikes of plants inoculated with BSMV:TaAG-A and BSMV:TaAG-B were wizened and the ripening rate was notably reduced.

To further verify the involvement of TaAG-A and TaAG-B in wheat male sterility, we assessed the relative transcript levels of the TaAG-A and TaAG-B genes in spikes by semi-quantitative RT-PCR analysis with tubulin as a control (Figure 5C). TaAG-A and TaAG-B were silenced in spikes after viral inoculation. At 16 and 20 days after inoculation, the relative transcript levels in spikes inoculated with RNA_γb and RNA_γb:GFP were decreased compared to the control, indicating that BSMV infection affected the growth and development of plants.