Lack of the α1,3-Fucosyltransferase Gene (Osfuct) Affects Anther Development and Pollen Viability in Rice

N-linked glycosylation is one of the key post-translational modifications. α1,3-Fucosyltransferase (OsFucT) is responsible for transferring α1,3-linked fucose residues to the glycoprotein N-glycan in plants. We characterized an Osfuct mutant that displayed pleiotropic developmental defects, such as impaired anther and pollen development, diminished growth, shorter plant height, fewer tillers, and shorter panicle length and internodes under field conditions. In addition, the anthers were curved, the pollen grains were shriveled, and pollen viability and pollen number per anther decreased dramatically in the mutant. Matrix-assisted laser desorption/ionization time-of-flight analyses of the N-glycans revealed that α1,3-fucose was lacking in the N-glycan structure of the mutant. Mutant complementation revealed that the phenotype was caused by loss of Osfuct function. Transcriptome profiling also showed that several genes essential for plant developmental processes were significantly altered in the mutant, including protein kinases, transcription factors, genes involved in metabolism, genes related to protein synthesis, and hypothetical proteins. Moreover, the mutant exhibited sensitivity to an increased concentration of salt. This study facilitates a further understanding of the function of genes mediating N-glycan modification and anther and pollen development in rice.


Introduction
N-glycosylation is a key post-translational modification that regulates the function of numerous proteins. N-glycoproteins have been implicated in diverse biological processes, such as protein folding, stability, protein-protein interactions, immune defense, inflammation, fertilization, embryogenesis and development; and also cause many diseases, including type 1 diabetes, Crohn's disease, rheumatoid arthritis, cancers, and neurological disorders [1,2]. N-glycoproteins alter the functions of numerous cell-surface proteins participating in adhesion and migration [3]. The initial N-glycosylation process occurs on the luminal side of the endoplasmic reticulum, where an oligosaccharide is assembled on the lipid carrier dolichylpyrophosphate and is transferred to the asparagine residue in a nascent

Identification and Isolation of a T-DNA Inserted Allele in the Rice Osfuct Gene
A T-DNA insertion line of mutant seeds was received from Kyung-Hee University, Republic of Korea [16]. The mutant seeds were grown under field conditions in 2012 (Suwon, Korea), and a phenotypic analysis was performed on this mutant line. The mutant displayed pleiotropic phenotypes, such as shorter height, fewer tillers and diminished shoot growth during all stages of development from seedling to maturity compared with the wild type ( Figure S1). Multiple database searches including those in the Rice Annotation Project Database (RAP-DB) (Available online: http://rapdb.dna.affrc. go.jp/), Basic Local Alignment Search Tool (BLAST) (URL) of the National Center for Biotechnology Information (NCBI; available online: https://blast.ncbi.nlm.nih.gov/Blast.cgi) and the Rice Functional Genomic Express Database (RiceGE) (Available online: http://signal.salk.edu/cgi-bin/RiceGE) of the Salk Institute Genomic Analysis Laboratory were performed using the rice Osfuct genomic DNA sequence as a query, and these analyses revealed a single copy of the gene present at the Os08g36840 locus in the rice genome. We have examined T-DNA insertion line based on screening for distorted segregation ratios by hygromycin selection. The mutant phenotype displayed a 3:1 ratio indicating that this mutant contained a single copy of the T-DNA. The predicted mRNA was 1542 bp in length and encoded α1,3-fucosyltransferase consisting of 513 amino acids ( Figure 1B,C).
Multiple sequence alignment of OsFucT with other plant glycosyltransferase proteins was performed using the ClustalW program (Available online: www.clustal.org/omega/), which revealed that α1,3-fucosyltransferase shared a conserved glycosyltrasnferase domain ( Figure S2). Five transmembrane prediction programs (MEMSAT, SOSUI, TMHMM, TSEG, DAS, and HMMTOP) were used to examine the possible transmembrane region of α1,3-fucosyltransferase. A single transmembrane domain was identified in α1,3-fucosyltransferase ( Figure 1C and Figure S2); the length of the transmembrane segments is underlined ( Figure S2). Hence, OsFucT contains a single transmembrane domain and a conserved glycosyltransferase motif. A phylogenetic analysis was carried out using MEGA6.0 (Available online: https://www.megasoftware.net/) [17] software and the neighbor joining method. The OsFucT-encoded protein shared 83% homology with those of Triticum aestivum and Hordeum vulgare ( Figure 1D). Therefore, these results suggest that Osfuct may have been generated from a common ancestor and the glycosyltransferase domain may have evolutionary and functional importance in plants. The mutant population was screened to identify the homozygous and heterozygous lines by polymerase chain reaction (PCR) analyses using gene-specific (P1, P2, and P3) and T-DNA border sequence primers (P4) ( Figure 1A,E). We identified a mutant allele with a T-DNA insertion in the Osfuct gene of rice. The T-DNA was added to the first intron of the gene in the mutant ( Figure 1A). The amplified PCR products were cloned and sequenced. No endogenous Osfuct transcript was detected in the homozygous mutant; however, the endogenous Osfuct transcript was observed in heterozygotes by reverse transcription-PCR (RT-PCR) analysis ( Figure 1F). Thus, these results suggest that T-DNA was added in the first intron of the gene and disrupted the gene open reading frame (ORF); therefore; the Osfuct transcript was absent in the homozygous mutant. The upper lane PCR products were amplified using gene-specific primers P1 and P3 (1.2 kb), whereas the lower lane PCR products were amplified using the gene-specific primer P3 and T-DNA border sequence primer P4 (0.6 kb). (F) Expression analysis of Osfuct in Dongjin (DJ), HM and HT lines was done using RT-PCR. The expression level of actin1 was used as a loading control and gene specific primers (P5 and P6) were used for RT-PCR analysis. The upper lane PCR products were amplified using gene-specific primers P1 and P3 (1.2 kb), whereas the lower lane PCR products were amplified using the gene-specific primer P3 and T-DNA border sequence primer P4 (0.6 kb). (F) Expression analysis of Osfuct in Dongjin (DJ), HM and HT lines was done using RT-PCR. The expression level of actin1 was used as a loading control and gene specific primers (P5 and P6) were used for RT-PCR analysis.

Osfuct Mutant Significantly Affects Anther Development
The Osfuct mutant exhibited a 30% reduction in plant height and fewer tillers compared to wild-type plants; however, the heterozygous plant showed a mild phenotype (Figure 2A). The Osfuct mutant had shorter panicles and reduced culm lengths than those of the wild type (Figure 2A,E-G). Furthermore, elongation of all internodes in the mutant was significantly inhibited compared with the wild type ( Figure 2B). Grain-filling was drastically reduced in the mutant and slightly affected in HT compared to Dongjin ( Figure 2C). Seeds were smaller in the mutant compared to Dongjin ( Figure 2D). Thus, these results demonstrate that α1,3-fucose transferase plays a critical role in vegetative growth and reproductive development in rice. Moreover, the mutant had a flower with a smaller panicle due to a defect in anther and pollen development, indicating that the Osfuct gene may have a flower development function in rice. To identify whether flower, pistil, anther, and pollen development are affected in the mutant, we examined the floral organs of mutant plants. Flowers of the wild type and mutant exhibited normal morphologies and development, including male and female organs; however, the mutant had smaller curved anthers ( Figure 3F,J) compared to the wild type ( Figure 3E,I). Heterozygous plants displayed less curved anthers ( Figure 3G,K). The length and width of mutant flowers, anthers and pistils were significantly decreased compared to heterozygous, rescue and wild type. The average mature flower lengths of homozygous, heterozygous, rescue, and wild-type plants were 6.12 ± 0.20, 6.29 ± 0.32, 6.45 ± 0.29, and 6.78 ± 0.28, respectively (  Figure 3M-P). The average ovule widths of homozygous, heterozygous, rescued and wild-type plants were 0.39 ± 0.02, 0.40 ± 0.03, 0.42 ± 0.01, and 0.38 ± 0.02 (Table 1, Figure 3M-P). Therefore, these results indicate that α1,3-fucose transferase plays an essential role in rice anther development.

Rescue of Mutant Phenotype by Wild-Type Osfuct Transgene
The mutation in the intron region of the Osfuct gene disrupted the ORF, which may impair the biological function of α1,3-fucosyltransferase. To assign a link between the T-DNA insertion and the mutant phenotype, we further confirmed this by a genetic complementation analysis using a genomic rescue construct. An 8066 kb genomic DNA fragment containing the entire Osfuct gene plus its 2852-bp promoter sequence was cloned into pCAMBIA3300 and transformed in the homozygous mutant ( Figure 3A). Transgenic rice plants were selected based on their resistance to L-phosphinothricin. The mutant and α1,3-fucose-deficient phenotypes were rescued in transgenic rice plants ( Figure 3B). Genotyping was done to confirm transgenic lines and further verified for the Osfuct transcript by RT-PCR analysis ( Figure 3C,D). The rescued lines also exhibited normal morphology and development of reproductive organs and recovered their anther morphology ( Figure 3H,L,P). These results show that the mutation in the intron region of the Osfuct gene is responsible for the pleiotropic phenotypes and demonstrate that the α1,3-fucose residue was added by Osfuct to the mutant ( Figure S3).

Effect of Osfuct Mutation on Pollen Development
To examine anther and pollen development in flowers, anther and pollen grains were collected from HM, HT, rescued, and wild-type plants. The pollen grains were stained with 80% (w/v) potassium iodide and 10% iodine. The wild type had more viable pollen grains that stained black indicating viable pollen ( Figure 4A,J). In contrast, the mutant had fewer viable pollen grains, and non-viable pollen grain stained orange ( Figure 4B,J). Furthermore, to investigate the defect in pollen grains, the ultra-surface structure of the pollen grains was examined by scanning electron microscopy (SEM) in HM, HT, rescued, and wild-type plants. The pollen grains were round and well developed in both the rescued and wild-type plants ( Figure 4E,H). However, the pollen grains were somewhat shriveled and significantly smaller in size in the HM line, as shown by the area of the pollen grain ( Figure 4F,K). We also counted the number of pollen grains per anther; the number was significantly lower in the HM line compared with rescued and wild-type plants ( Figure 4I). These results demonstrate that the mutation affects pollen morphology, viability, and pollen number per anther in rice.
well developed in both the rescued and wild-type plants ( Figure 4E,H). However, the pollen grains were somewhat shriveled and significantly smaller in size in the HM line, as shown by the area of the pollen grain ( Figure 4F,K). We also counted the number of pollen grains per anther; the number was significantly lower in the HM line compared with rescued and wild-type plants ( Figure 4I). These results demonstrate that the mutation affects pollen morphology, viability, and pollen number per anther in rice. Data are mean ± SD (n ≥ 6). Asterisk indicates significant differences compared with wild type (* p < 0.05 by Student's t-test).

The Glycosylation Pattern Is Altered in the Osfuct Mutant
We performed multiple sequence alignment of α1,3-fucosyltransferase amino acid sequences with homologs from other plant species. The OsFucT homologous protein sequences were retrieved from the National Center for Biotechnology Information database (Available online: http://www. ncbi.nlm.nih.gov/). The alignments revealed that α1,3-fucosyltransferase shared a highly conserved glycol transferase domain. α1,3-Fucosyltransferases are responsible for transferring the α1,3-linked fucose residue to the glycoprotein N-glycan. To investigate the glycosylation pattern, HM, HT, rescued and wild-type plants were subjected to total N-glycan analysis by mass spectrometry. Lack of fucose residues in the N-glycan mutant was examined by subtracting 146 mass units from rescued and wild-type plants. The mass spectra derived from the mutant were significantly different from those of the wild-type and rescued lines ( Figure S3A-D). The fucose residue was missing from the N-glycan structure of the mutant. The four major peaks of the complex-type N-glycans differed in the mutant and wild type ( Figure S3E, Table S3 . These results suggest that inactivation of the Osfuct gene in the mutant produced N-glycans without α1,3-fucose residues, whereas the rescued lines were capable of producing the complex N-glycans including the α1,3-fucose residue.

Osfuct Mutant Exhibits Sensitivity to Salt Stress
We performed shoot growth assays to investigate salt tolerance in the mutant on MS medium supplemented with 0, 50, 100, 150, and 200 mM NaCl ( Figure 5A,B). However, the mutant plants were smaller compared to the wild type. Growth was monitored for 2 weeks, and the mutant exhibited sensitivity to salt stress and overall growth in comparison to the wild type. The mutant plants exhibited pale-green leaves phenotype compared to the wild type. The plants were most sensitive to the highest concentration of NaCl (200 mM). The most sensitive phenotype was evaluated in term of shoot growth over 14 days, and all mutant lines showed sensitivity and poor growth at the highest concentration of NaCl (200 mM). The fresh weight of wild-type shoots was measured after 2 weeks of growth at 0, 50 Figure 5A,B). In addition, a positive correlation was found between increase salt concentration and proportionally reduction of fresh weight of mutant plants. These results suggest that this rice mutant is sensitive to high salt concentrations. Seedlings of the Dongjin (DJ) and mutant lines (HM) were exposed to a nutrient solution containing 0-200 mM NaCl. After 3 weeks, individual leaves were separated and shoot length (A) and weight were measured (B). Data are mean ± SD (n = 15). Asterisk indicates a significant difference compared with wild type (* p < 0.05 by t-test). Error bars show standard deviation.

Transcriptomic Profiling of the Osfuct Mutant
We performed a whole transcriptome analysis of the mutant (HM) and wild type (Dongjin) to identify the possible underlying molecular mechanisms that could explain the pleiotropic functions of Osfuct during growth and reproductive development. To investigate the effect of Osfuct mutation on transcriptional regulation of other pathway genes, whole transcriptomic profiling was performed with three biological replicates in mutant and wild type (Dongjin) plants, respectively. The rice whole genome 135 K oligo microarray (GreenGene Biotech., Inc., Seoul, Korea) was used for the transcriptome profiling ( Figure S4). The transcriptome analysis and data normalization were carried out as described in the Materials and Methods section ( Figures S5 and S6, Tables S6 and S7). Differentially expressed (either up or downregulated) transcripts between the mutant and wild type were determined via log2 fold change above 2.0 (p < 0.05) and log2 fold change above 1.6 (p < 0.1) change thresholds. Among the 42 significantly differentially expressed genes, 32 were upregulated (Tables 2 and S4), and 10 were downregulated (Tables 2 and S5). We categorized those genes into Seedlings of the Dongjin (DJ) and mutant lines (HM) were exposed to a nutrient solution containing 0-200 mM NaCl. After 3 weeks, individual leaves were separated and shoot length (A) and weight were measured (B). Data are mean ± SD (n = 15). Asterisk indicates a significant difference compared with wild type (* p < 0.05 by t-test). Error bars show standard deviation.

Transcriptomic Profiling of the Osfuct Mutant
We performed a whole transcriptome analysis of the mutant (HM) and wild type (Dongjin) to identify the possible underlying molecular mechanisms that could explain the pleiotropic functions of Osfuct during growth and reproductive development. To investigate the effect of Osfuct mutation on transcriptional regulation of other pathway genes, whole transcriptomic profiling was performed with three biological replicates in mutant and wild type (Dongjin) plants, respectively. The rice whole genome 135 K oligo microarray (GreenGene Biotech., Inc., Seoul, Korea) was used for the transcriptome profiling ( Figure S4). The transcriptome analysis and data normalization were carried out as described in the Materials and Methods section ( Figures S5 and S6, Tables S6 and S7). Differentially expressed (either up or downregulated) transcripts between the mutant and wild type were determined via log 2 fold change above 2.0 (p < 0.05) and log 2 fold change above 1.6 (p < 0.1) change thresholds. Among the 42 significantly differentially expressed genes, 32 were upregulated (Table 2 and Table S4), and 10 were downregulated ( Table 2 and Table S5). We categorized those genes into two groups to make the broad knowledge and interpret our microarray data clearly. Group 1 genes included those with the log 2 fold change above 2 expression change threshold (p < 0.05), while group 2 genes included those with the log 2 fold change above 1.6 threshold (p < 0.1). The expression of several key genes involved in plant growth and development was altered in the mutant compared with the wild type ( Table 2, Tables S4 and S5). Protein kinases, transcriptional factors, metabolism-related genes, and hypothetical genes comprised a significant portion of the differentially expressed genes in the mutant compared with the wild type. The following genes were significantly upregulated: putative disease resistance protein RGA4 required to recognize Avr (AVR1-CO39) to mediate disease resistance [18]; LRR (Leucine-rich repeat)receptor-like serine/threonine-protein kinase regulating the diverse developmental processes [19]; putative wall-associated protein kinase involved in cell expansion and response to external stimuli [20]; F-box protein-like implicated in phytohormone signaling [21]; isocitrate lyase involved in glyoxylate pathway [22]; disheveled-associated activator of morphogenesis 1 involved in regulating the actin cytoskeleton [23]; glutathione S-transferase implicated in abiotic tolerance [24]; transcription factor (basic helix-loop-helix) bHLH35 regulating diverse biological processes including epidermal differentiation, seed germination, carpel and anther development, response to phytochrome, diverse biotic and abiotic stresses [25]; Dof zinc finger protein involved in wide range of fundamental processes in higher plants [26]; transcription initiation factor IIE involved in gene expression [27]; eukaryotic translation initiation factor 4A-3 required for protein synthesis [28]; and hypothetical proteins. Significantly downregulated genes included glycoprotein 3-α-L-fucosyltransferase A function in N-glycosylation [1,2]; the WASH complex subunit CCDC53 homolog regulating actin polymerization [29]; kinesin-like protein (KIF19) and cortical cell-delineating protein involved in morphogenesis, mitosis, and signal transduction [30]; endosperm defective 1 (EDD1) required for microtubule function [31]; glycine-rich RNA-binding protein 2 play crucial roles in post-transcriptional gene regulation [32]; L-type lectin-domain containing receptor kinase in plant developmental processes including signal transduction, response to hormone, biotic and abiotic stresses [33]; non-specific lipid transfer protein implicated in key cellular processes including cell wall organization, the stabilization of membranes, and signal transduction [34]; and some hypothetical genes.
To further verify whether the transcript changes identified by the microarray analysis were reliable, total RNA from the same mutant and wild type was subjected to quantitative RT-PCR (qRT-PCR) analyses. The 42 genes that were either significantly up-or down-regulated in the microarray analysis were selected for qRT-PCR analysis ( Table 2, Tables S4 and S5). We observed a significant correlation between the qRT-PCR and microarray results. Although some differences were found between the microarray and qRT-PCR analysis, the pattern of differentially expressed genes detected by the two approaches remained consistent. Thirty-two (72%) of the 42 genes corresponded to the microarray results, and the expression of 10 genes varied in both approaches ( Figure 6 and Figure S7). Based on the earlier reported functions of these up-and downregulated genes [18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34] we proposed a possible working model in Figure 7 to depict the underlying molecular mechanism of Osfuct function in the pleiotropic development defects in rice. This model demonstrates that loss of function Osfuct modulates the genes involved in plant developmental processes, such as transcriptional factors and protein kinases, which affect growth as well as anther and pollen development in later stages. (Figures 6 and S7). Based on the earlier reported functions of these up-and downregulated genes [18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34] we proposed a possible working model in Figure 7 to depict the underlying molecular mechanism of Osfuct function in the pleiotropic development defects in rice. This model demonstrates that loss of function Osfuct modulates the genes involved in plant developmental processes, such as transcriptional factors and protein kinases, which affect growth as well as anther and pollen development in later stages.

Discussion
In this study, we demonstrated that loss of Osfuct dramatically impeded growth and flower development in rice. To our knowledge, this is the first report on the function of Osfuctin anther and pollen development in the Osfuct mutant. In addition, the anther was curved in the HM mutant ( Figure 3F,J), while the anther was less curved and showed a milder phenotype including plant height, tiller number, panicle lengths and elongation in the HT plant ( Figure 3G,K) compared with the wild type ( Figure 3E,I). A single copy of the functional Osfuct allele may compensate for partial function in the HT plant. We provide strong evidence that loss of Osfuct impairs pollen and anther development in the mutant.

Discussion
In this study, we demonstrated that loss of Osfuct dramatically impeded growth and flower development in rice. To our knowledge, this is the first report on the function of Osfuct in anther and pollen development in the Osfuct mutant. In addition, the anther was curved in the HM mutant ( Figure 3F,J), while the anther was less curved and showed a milder phenotype including plant height, tiller number, panicle lengths and elongation in the HT plant ( Figure 3G,K) compared with the wild type ( Figure 3E,I). A single copy of the functional Osfuct allele may compensate for partial function in the HT plant. We provide strong evidence that loss of Osfuct impairs pollen and anther development in the mutant.
Microarray analyses revealed that several genes essential in plant developmental processes were significantly altered in Osfuct mutant compared to wild-type. Receptor-like protein KINASE2 (RPK2) T-DNA insertional mutants exhibit defects in anther and pollen development in Arabidopsis [35]. LRR-RLK and MSP1 (Multiple Sporocyte1) determine anther cell identity in rice [36]. Similarly, downregulation of Embryonic flower 2b (OsEMF2b) causes defects in anther and pollen development in rice [37]. Our microarray results show that the expression of leucine-rich repeat receptor-like kinase (LRR-RLK) was significantly elevated in the mutant ( Table 2). bHLH transcription factor called Dysfunctional Tapetum1 (DYT1) is essential for anther development in Arabidopsis [38]; the transcript level of bHLH transcription factor was also significantly up-regulated in the mutant. There are several studies showing that (gibberellic acid) GA synthesis and signaling are important for anther and pollen development. Arabidopsis loss-of-function mutant in copalyl diphosphate synthase (ga1) has poorly developed anthers and pollen grains also are not viable [39]. Our results are agreement with the previous findings and suggest that Osfuct may affect anther and pollen development processes by altering the expression of genes implicated in gametophyte development in rice. However, how Osfuct participates in pollen and anther development remains elusive ( Figure 3E-L and Figure 4A-K).
The Osfuct mutant displayed pleiotropic developmental defects, such as shorter plant height, fewer tillers, shorter panicle length and internode elongation (Figure 2A,B and Figure 3B). We recorded all phenotypic data from the field-grown plants to avoid the conditional phenotypic variation. MYB-like gene of plant height 1 (OsMPH1) has been implicated in the regulation of plant height through change internode cell length in rice. Overexpression of OsMPH1 leads to increases of plant height and grain yield, while knockdown of OsMPH1 showed the opposite phenotypes in rice. RNA-sequencing data also revealed that expression profile of cell elongation and cell wall synthesis-related genes significantly modulated in transgenic rice plants [40]. Dwarf Tiller1 (DWT1) regulates the shoot and tiller formation in rice. Mutant plants had main shoots with normal height and larger panicles but dwarf tillers had smaller panicles compared to the wild-type. The DWT1 gene is highly expressed in young panicles. Mutant was defective in internode elongation is associated with gibberellin signaling [41]. Rice ethylene-response AP2/ERF (OsEATB) factor regulate the internode elongation by repressing a gibberellin biosynthetic gene, ent-kaurene synthase A [42]. Our results also consistent with these findings and indicate that Osfuct may possess important biological and developmental functions in rice. The expression level of DNA-binding one zinc finger (DOF) in the Osfuct mutant increased significantly as well; the DOF transcription factor family has been implicated in a wide range of fundamental processes in higher plants, including seed germination, maturation, photosynthesis, flower induction, responses to phytochrome, plant hormone signaling, and biotic and abiotic stressors [26]. Overexpression of OsDof12 in rice reduces height, shortens leaf length, produces more erect leaves and smaller panicles, and decreases grain yield in transgenic rice plants [43]. Collectively, these results demonstrated that loss of Osfuct may modulate the expression of genes participating in plant growth and developmental processes.
In addition, the transcript levels of the F-box proteins increased significantly in the mutant compared to wild type through microarray data; F-box proteins have also been implicated in phytohormone signaling through ubiquitin-mediated protein degradation [44]. An optimal concentration of phytohormones is essential for various developmental processes, such as embryogenesis, tissue differentiation, and organogenesis. The fuct-1 and fuct-2 mutants affect basipetal auxin transport and accumulation resulting in decreased gravitropic responses in rice [15]. Our results were consistent with earlier findings that the lack of fucose residues in the mutant may affect the function of the key genes involved in phytohormone signaling and plant developmental processes, which subsequently may affect growth and development in the mutant [15]. However, how Osfuct regulates the underlying molecular mechanisms of plant developmental processes remains to be established. Furthermore, the pollen grains were shrunken and significantly smaller in size in the mutant compared to the restored and wild type. Additionally, pollen viability and pollen number per anther decreased dramatically in the mutant ( Figure 4A-K). The mutant seeds were smaller and the total number of seeds per plant also decreased. Pollen development is also defective in gibberellin-deficient mutants gib-1 and gib-2 in the tomato [45]. Lalanne et al. [9] demonstrated that a defect in SETH1 and SETH2 affects pollen germination, tube growth, and male transmission in Arabidopsis. SETH1 and SETH2 are involved in the first step of the glycosylphosphatidylinositol biosynthetic pathway. Mutation in two phytochrome genes (PHYA and PHYB) affect the anther and pollen development in rice [46]. Isocitrate lyase is a key enzyme in the glyoxylate pathway that is widespread in nature and functions in diverse physiological processes in many organisms [22]. Isocitrate lyase and malate synthase genes from Brassica napus are active at specific stages of pollen development [22]. Our microarray analyses demonstrated that the isocitrate lyase transcript level was significantly induced in the mutant. Loss of Osfuct may affect expression of these genes and also deregulate genes that play an essential role in anther and pollen development. Further, the transcript level of kinesin-like protein KIF19 decreased significantly as wellin the mutant. AtNACK1 and AtNACK2 are members of the kinesin-7 family essential for cell-plate formation in pollen and male and female gametogenesis [47]. Similarly, the pollen semi-sterility 1 (PSS1) mutant encodes a kinesin-1-like protein affecting spikelet fertility with reduced pollen viability and defective anther dehiscence in rice [48]. These results indicate that loss of Osfuct may result in aberrant regulation of the expression of genes implicated in anther and pollen development. However, regulation of pollen development processes by Osfuct remains unclear. To verify the mutant phenotype, we conducted a genetic complementation test and introduced the entire Osfuct gene, including the upstream region, into the mutant. Genetic complementation analyses revealed that the transgenic rice lines restored the pleiotropic phenotype of mutant (Table 1; Figures 3B and 4D,H-K); hence, the mutation in the intron region of Osfuct was responsible for the diverse developmental abnormalities. Two rice mutants (fuct-1 and fuct-2) with loss of Osfuct function displayed larger tiller angles, shorter panicle lengths and internodes, reduced grain filling, and an increase in the number of chappy grains with unusual shapes [15]. They also performed a complementation analysis and mobilized the cDNA with a promoter containing a portion of the first exon and restored the phenotype. Taken together, these results suggest that Osfuct may have pleiotropic functions in various developmental processes in rice.
N-glycosylation plays an important role in regulating protein folding, subcellular localization, and function of secreted proteins. Mutants interrupted in N-glycosylation biosynthesis have impaired reproduction and embryo formation [50]. OsDGL1 homologs of the oligosaccharyltransferase complex subunit have been implicated in N-glycosylation and mutation in Osdgl1, causing a defect in N-glycosylation in rice. The Osdgl1 mutant modulates root cell wall polysaccharide composition, leads to a smaller root meristem, shorter root cells, and root cell death [8]. α1,3-Fucosyltransferase and β1,2-xylosyltransferase are deactivated by multiplex CRISPR/Cas9 in Nicotiana tabacum BY-2 cells that produce glycoproteins deficient in plant-specific N-glycans [14]. Our results also demonstrate that the mutant exhibited pleiotropic phenotypic defects, including impaired growth, fewer tillers, shorter plant height, panicle, culm, impeded internode elongation, and defects in anther and pollen development under field conditions ( Figure 2). These results indicate that missing fucose residues in the N-glycan complex in the mutant affected the expression of genes that play a crucial role in plant developmental processes. How N-glycosylation affects vegetative and reproductive development remains to be determined. Moreover, the Osfuct mutant was sensitive and showed poor growth at increased salt concentration. As a previous study showed, that defect in N-glycan synthesis, processing, and maturation of complex glycan 1 (cgl1) makes the plants more salt-sensitive compared with the wild type [11]. Thus, the proper maturation of N-glycan in the Golgi complex is essential for salt tolerance in Arabidopsis [14]. Arabidopsis T-DNA insertion mutant (alg10-1) encoding α1,2-glucosyltransferase exhibits a severe defect in N-glycosylation, leaf growth, and increased salt sensitivity [13]. Transcriptome results also show that the glutathione S-transferase expression was upregulated more than threefold in the mutant. Glutathione S-transferases are crucial detoxification enzymes involved in abiotic stress tolerance by metabolizing numerous toxic compounds through glutathione conjugation and play a critical role improving tolerance to cold, osmotic dehydration, salt, and herbicide-induced damage [24]. Ectopic expression of glutathione S-transferase enhances salt and drought tolerance in Arabidopsis [51]. The transcript level of lectin receptor-like kinase (LecRLK) decreased significantly in the mutant. LecRLK is known to play an important role in plant developmental processes, including signal transduction and responses to hormone, biotic, and abiotic stressors [33]. The expression of non-specific lipid-transfer proteins (nsLTPs)_was significantly repressed in the Osfuct mutant. nsLTPs are also play an essential role in plant growth and development, sexual reproduction, seed germination and development, tolerance to biotic and abiotic stress [34]. Our results also show that the Osfuct mutant was sensitive to the highest concentration of NaCl (200 mM) and that the fresh weight of shoots was proportionally reduced compared to wild-type plants ( Figure 5A,B). Therefore, it is intriguing to assume that a deficiency of α1,3-fucose may affect key signaling pathways and the expression of genes related to the abiotic stress response in rice. However, it is unknown how OsFucT regulates the abiotic stress response in the mutant via N-glycosylation. Finally, expression of 19 hypothetical genes was dramatically induced/repressed (induced 17; repressed 2) in the mutant; thus, it is impossible to speculate about their possible effects on rice growth and development at present.
In summary, we characterized a novel Osfuct mutation that significantly affects anther and pollen development in rice. The loss of Osfuct caused pleiotropic morphological abnormalities, such as shorter plants, reduced panicle length, inhibited internode elongation, and impaired anther and pollen development. Several genes involved in growth and development processes were also drastically altered in the mutant. Thus, collectively our results demonstrate that Osfuct plays an essential role in anther and pollen development and has potential applications in a rice-breeding program for producing humanized N-glycan proteins. Several questions remain to be addressed. How does Osfuct act at the cellular level? Identification and characterization of Osfuct-interacting partners to dissect the intricate regulatory gene network may shed further insight into the functions of Osfuct impaired growth, anther, and pollen development.

Plant Materials and Growth Conditions for Mutant Screening
The T-DNA insertion line of mutant seeds (PFG_2A-30078.L) was obtained from Kyung-Hee University, Republic of Korea [19]. The rice (Oryza sativa cv Dongjin) and mutant seeds were sterilized by dipping in 70% ethanol for 1 min and washed three times with double-distilled water. The seeds were sterilized repeatedly with 2% sodium hypochlorite for 30 min, washed five times with double-distilled water, and germinated on 2N6 media in a growth chamber under a 16-h light (28 • C)/8-h dark (26 • C) photoperiod during the full life cycle. The seedlings were transferred to individual pots and grown under field conditions during the 2012 season (Suwon, Korea) to measure the morphological phenotypes of rice.

Genotyping and Phylogenic Analysis of the Osfuct Mutant
T-DNA insertion and homozygosity were verified by genomic PCR using primers specific to Osfuct and T-DNA. Genomic DNA was extracted from leaves of 4-week-old plants according to the CTAB protocol [52]. Genomic DNA purity and concentration were measured using a NanoDrop spectrophotometer (ND-1000; Nanodrop Technologies, Wilmington, DE, USA). T-DNA insertion and homozygosity of Osfuct were confirmed by the gene-specific primer pair (P1 + P3), the T-DNA left border, and the gene-specific primer (P4 + P3) used in the PCR (Biometra TProfessional ThermoCycler, Gottingen, Germany; Phire ® Hot Start DNA Polymerase, Finnzymes, Ipswich, MA, USA). The PCR was carried out with the initial reaction of 98 • C for 5 min; followed by 35 cycles of 98 • C for 5 s, 67 • C for 5 s, and 72 • C for 1 min; with a final reaction of 1 min at 72 • C. All primers used in the analysis are presented in Table S1.

Reverse Transcription-Polymerase Chain Reaction (RT-PCR) and Quantitative Real-Time PCR (qPCR) Analysis
Total RNA from frozen rice leaves was extracted in liquid nitrogen [53]. Briefly, frozen rice leaves were ground to powder and transferred to a microcentrifuge tube. A total of 800 µL of extraction buffer (200 mM Tris-Cl, 400 mM LiCl, 25 mM EDTA, and 1% SDS, pH 9.0) and 600 µL of acidic phenol was added, vortexed for 10 s, and the solution was kept on ice for 30 min followed by centrifugation at 10,000× g for 2 min at 4 • C. The supernatant was transferred to a microcentrifuge tube and 600 µL phenol and chloroform were added and centrifuged at 10,000× g for 5 min at 4 • C. The supernatant was transferred to a new microcentrifuge tube and precipitated in 250 µL of 8 M LiCl at −70 • C overnight. After centrifugation, the pellets were washed twice with 70% cold ethanol, completely dried, and dissolved in (Diethyl pyrocarbonate) DEPC-treated distilled water. Genomic DNA contamination was removed by treating with DNase I (Takara Bio. Inc., Shiga, Japan) at 37 • C for 30 min and cDNA was generated by reverse transcribing the mRNA using the iScript TM cDNA synthesis kit (Bio-Rad, Hercules, CA, USA) at 46 • C for 20 min. cDNA was quantified using a spectrophotometer (ND-1000), and subjected to RT-PCR analysis using the primer set (P5 + P6) with the following PCR profile: 40 cycles of 98 • C for 10 s, 55 • C for 5 s, and 72 • C for 1 min. The actin1 (XM_015774830) primer was used as an internal loading control. Amplified RT-PCR products were further confirmed by DNA nucleotide sequencing. Quantitative real-time PCR (qPCR) was performed using a CFX96 Real-Time PCR Detection System (Bio-Rad) with the SYBR Premix (Toyobo, Osaka, Japan). The qPCR reactions were run with the following profile: denaturation at 95 • C for 1 min, followed by 39 cycles of denaturation at 95 • C for 15 s, and annealing at 55-60 • C for 30 s. A melting curve was obtained through a protocol of 95 • C for 1 min, 95 • C for 15 s, 65 • C for 1 min followed by cooling at 40 • C for 10 min. qPCR was performed with three technical replicates and repeated three times for each gene. CFX Manager 2.0 (Bio-Rad) was used to calculate the cycle threshold value. Rice ubiquitin1 (OsUBQ1, XM_015774309) was used as the control to normalize the data. Primer pairs were designed using PrimerQuest Tool (Integrated DNA Technologies, Coralville, IA, USA) and all primers used in this study are listed in Table S2.

Phenotypic Analyses of the Osfuct Mutant
Seeds of Osfuct and wild-type rice were sown and transferred to pots to grow under field conditions following the full lifecycle for a phenotypic examination during different developmental stages. To evaluate morphological phenotypes including plant height, tiller number, shoot, culm, panicle, and internode elongation length, the mutant and wild-type (Dongjin) rice were grown under field conditions during the 2012 season (Suwon, Korea) to follow the full lifecycle. The mutant and wild-type (Dongjin) rice were grown in greenhouse conditions to analyze reproductive development parameters, such as length and width of mature flowers, anthers, pistils, pollen grain viability, and pollen grains per anther.

Assessment of Reproductive Development of the Osfuct Mutant
Flowers were collected from mutant and wild-type rice plants to analyze reproductive development. The anther and pistil images were taken with a microscope (SteREO Discovery V12, Carl Zeiss, Zena, Germany). To examine the ultrastructure of pollen, Osfuct (HM, homozygote, HT, heterozygote), rescued, and wild-type pollen images were captured by a JEOL JSM-7500FA (JEOL Ltd., Tokyo, Japan) analytical field-emission scanning electron microscope after coating with tungsten particles using ion (magnetron) sputter coater (Hitachi E-1030) in vacuum. The pollen grains were stained with 80% (w/v) potassium iodide and 10% iodine to determine pollen viability.

Vector Construction for Complementation Analysis of Osfuct Disrupted Mutant
An 8066-bp fragment of genomic DNA plus a 2852-bp upstream sequence that contained the promoter region cloned into the SmaI site of the binary vector pCAMBIA3300 after modification of restriction enzyme sites were used for the complementation analysis of the mutant. The promoter, genomic, and cDNA fragments were amplified using the primer pairs (P7 + P6) and (P5 + P6) and the PCR profile described above. Nucleotide sequences of the cloned fragments were verified by an automated DNA sequencer (3730xI DNA Analyzer, Applied Biosystems, Foster City, CA, USA). These plasmid constructs were electroporated into the Agrobacterium tumefaciens strain LBA4404 and introduced into the homozygote mutant by A. tumefaciens-mediated transformation with some modifications as described previously [54]. Transgenic rice plants were selected on L-phosphinothricin (6 mg/L) containing medium. L-phosphinothricin resistance transgenic rice plants were transferred to soil and allowed to grow in a greenhouse, as described above, for further phenotypic analysis.

N-Glycan Profiling of the Osfuct Mutant
N-glycans were prepared and purified from the mutant and wild-type rice leaves according to Karg et al. [55]. The N-glycan analysis was performed in positive-ion reflectron mode using a matrix-assisted laser desorption ionization-time of flight mass spectrometer (MALDI-TOF MS) with an Autoflex system from Bruker Daltonics (Bruker, Billerica, MA, USA). The tandem mass spectrometric analysis was carried out using an Axima Resonance MALDI-quadrupole ion trap-TOF instrument (Shimadzu, London, UK). 5-Dihydroxybenzoic acid (10 mg/mL in 50% methanol) was used as the matrix. The data analysis was performed with Flex Analysis 3.3 software (Bruker) and Launchpad 2.9.3 software (Kratos Analytical Ltd., Cambridge, UK). The calculated mass of the major N-glycans was compared with earlier reports to assign the corresponding structures [12,13,56].

Salt Tolerance Assay of the Osfuct Mutant
Young mutant plants were subjected to high salinity stress assay. The young mutant and wild-type rice plants were grown in 0, 50, 100, 150, and 200 mM NaCl-containing medium for 2 weeks. Shoot length and weight were scored, and the plants were photographed. Four biological replicates of rice plants were used for the salt tolerance assay.

Transcriptome Analyses of the Osfuct Mutant
Seeds of mutant (homozygote; HM) and wild-type (Dongjin) rice were sterilized and germinated as previously described in Section 4.1 in the growth chamber. Seedlings were transferred to individual pots and grown in a greenhouse. Three biological replicates of HM and Dongjin rice plants were selected respectively for transcriptome analysis using the rice whole genome 135 K oligo microarray (GreenGene Biotech., Inc., Seoul, Korea). These microarrays contained 31,439 genes submitted to the International Rice Genome Sequencing Project (IRGSP, available online: http://rgp.dna.affrc.go.jp/E/ IRGSP/) and the Rice Annotation Project version 2 (RAP2, available online: http://rapdb.dna.affrc.go. jp/). Total RNA was extracted from leaves of 4-week-old plants at the tillering stage, and genomic DNA was removed by DNase I (Takara Bio) digestion. Double-stranded cDNAs were synthesized using the RevertAid TM H Minus First Strand cDNA Synthesis Kit (Fermentas, Vilnius, Lithuania) according to the manufacturer's directions. Briefly, total RNA (10 µg) was mixed with following components (4 µL of 5× first strand buffer, 1 µL of oligo dT primer (100 µM), 1 µL RiboLock TM Ribonuclease inhibitor, 2 µL of 10 mM deoxynucleotide mix and 1 µL of RevertAidTM H Minus M-MuLV Reverse transcriptase enzyme) and incubated at 42 • C for 1 h. Reverse transcription and (Double-stranded DNA) dsDNA synthesis were terminated by incubating at 70 • C for 10 min. Cy3-labeled DNAs were synthesized by adding the following master mix (1 µg of double-stranded cDNA, 30 µL of Cy3-9mer primers, 2 µL of Klenow fragments (50 U/µL, Takara), 10 µL of 50× deoxynucleotide mix (10 mM each), and 8 µL of double distilled water) and incubated at 37 • C for 2 h. The labeled dsDNA was purified using the MinElute Reaction Cleanup Kit (Qiagen, Valencia, CA, USA) and hybridization was carried out with the MAUI chamber (Biomicro, Salt Lake City, UT, USA) at 42 • C for 16-18 h. After hybridization, the microarrays were washed at 42 • C for 10-15 s with GE Wash Buffer I, again for 2 min at 42 • C with GE Wash buffer II, and finally washed with Wash buffer III for 1 min with agitation. The microarray was dried immediately by centrifugation at 500× g for 1 min and scanned using GenePiX scanner 4000 B (Molecular Devices, Inc., Sunnyvale, CA, USA). The signal was digitized and analyzed by Nimblescan (Madison, WI, USA). To increase the sensitivity and reproducibility of the transcriptome analysis, the data were normalized using the cubic spline normalization method, and probe-level summarization was processed by Robust Multichip Analysis using a median polish algorithm. Fold-change was calculated using Acuity 3.1 (Axon Instruments, Sunnyvale, CA, USA). Greater than two-or log 2 fold change above 1.6 with p < 0.05 (or 0.1) was set as the threshold for statistical significance. The Osfuct mutant gene expression profiles have been deposited in NCBI's Gene Expression Omnibus and are accessible through Gene Expression Omnibus (GEO) Series accession number GSE110873 (Available online: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE110873).