Genome Wide Identification and Characterization of Wheat GH9 Genes Reveals Their Roles in Pollen Development and Anther Dehiscence

Glycoside hydrolase family 9 (GH9) is a key member of the hydrolase family in the process of cellulose synthesis and hydrolysis, playing important roles in plant growth and development. In this study, we investigated the phenotypic characteristics and gene expression involved in pollen fertility conversion and anther dehiscence from a genomewide level. In total, 74 wheat GH9 genes (TaGH9s) were identified, which were classified into Class A, Class B and Class C and unevenly distributed on chromosomes. We also investigated the gene duplication and reveled that fragments and tandem repeats contributed to the amplification of TaGH9s. TaGH9s had abundant hormone-responsive elements and light-responsive elements, involving JA–ABA crosstalk to regulate anther development. Ten TaGH9s, which highly expressed stamen tissue, were selected to further validate their function in pollen fertility conversion and anther dehiscence. Based on the cell phenotype and the results of the scanning electron microscope at the anther dehiscence period, we found that seven TaGH9s may target miRNAs, including some known miRNAs (miR164 and miR398), regulate the level of cellulose by light and phytohormone and play important roles in pollen fertility and anther dehiscence. Finally, we proposed a hypothesis model to reveal the regulation pathway of TaGH9 on fertility conversion and anther dehiscence. Our study provides valuable insights into the GH9 family in explaining the male sterility mechanism of the wheat photo-thermo-sensitive genetic male sterile (PTGMS) line and generates useful male sterile resources for improving wheat hybrid breeding.


Introduction
Glycoside hydrolase family 9 (GH9) encodes a cellulase gene consisting of endo-β-1,4glucanase (EGase) [1], which could hydrolyze soluble cellulose into reducing oligosaccharides. In addition, it was also shown that the GH9 family could hydrolyze cellulose-related polysaccharides such as carboxymethyl (CM) cellulose, phosphate-swollen cellulose and 1,3-β-1,4-β-glucan [2]. EGases have been studied in many species, such as eukaryotes and prokaryotes [3,4]. Based on their binding sites, the GH9 family can be divided into three subfamilies: class A (GH9A), class B (GH9B) and class C (GH9C) [5]. It has been reported that the members of GH9A and GH9B are involved in cell elongation and fruit maturation and abscission. GH9C has a cellulose-binding structural domain localized on the plasma membrane carbohydrate binding modules (CBM49). All these GH9s have been reported to be associated with cellulose biosynthesis [6].

Chromosomal Localization of the TaGH9 Family and Synteny Analysis
To understand the relative positions of the GH9 genes on wheat chromosomes, we tagged their relative positions on wheat chromosomes. A total of 74 copies of TaGH9 were distributed on 20 chromosomes, except chromosome 3B. The distribution of genes on chromosomes was scattered and there was no cluster distribution ( Figure S1).
The synteny analysis revealed that there were three TaGH9s occurring in tandem

Chromosomal Localization of the TaGH9 Family and Synteny Analysis
To understand the relative positions of the GH9 genes on wheat chromosomes, we tagged their relative positions on wheat chromosomes. A total of 74 copies of TaGH9 were distributed on 20 chromosomes, except chromosome 3B. The distribution of genes on chromosomes was scattered and there was no cluster distribution ( Figure S1).

Structure Analysis and Motif Distribution of the TaGH9 Family
The gene structure of TaGH9 genes was understood by analyzing the exon-intron distribution. As indicated in Figure S3 (left), TaGH9 contained two to seven exons and introns. It was found that copies of the same TaGH9 genes had similar or identical intronexon constructions. For example, TaGH9-13 have the same intron-exon structures among their copies, while the copies of TaGH9-19 and TaGH9-9 in a different subclade display different structures.
We also investigated the motifs of TaGH9 to further revel the role of TaGH9 using the MEME tool. In this study, ten types of motifs were charactered. The results show that the number of conserved motifs for each TaGH9 protein is mostly between seven and ten. Here, we found that TaGH9-6, TaGH9-9 and TaGH9-10 have 9 motifs without motif 9 ( Figure S3 (right)). The results show that the exon-intron structures and motifs of different TaGH9 members are different. However, the exon-intron structures and motifs were similar or the same in the same TaGH9 subfamily. This suggests that the functional divergence of different TaGH9s may be caused by different exon-intron and motifs.

Cis-Acting Elements in Promoter Have Valuable Information for Analysis of the TaGH9 Family Function
The cis-acting element plays a key role in gene expression as an essential component of the gene promoter region, which assists to understand how the gene responds to external stimulation [30]. As shown in Figure 2, 29 categories of cis-element were identified, including light response-related elements, plant growth and development-related elements, hormone-response-related elements and biotic/abiotic adversity stress-related elements ( Figure 2 and Table S2). G-box, a light-response-related element, GARE-motif and TC-rich elements with temperature-related TGACG-motif, a MeJA-response elements and ABRE (ABA responsive element), an abscisic acid-response-related element, were found in the TaGH9 gene family. There are differences in the cis-acting elements between different TaGH9s. For example, TaGH9-11 and TaGH9-12 had more light-responsive elements and hormone-responsive elements than stress-related elements and growth and developmentrelated elements ( Figure 2). LTR (low-temperature response) elements concerned with low-temperature treatment were enriched in TaGH9-11, suggesting that the TaGH9 family has different roles since it owns different cis-acting elements.

TaGH9 Genes Were Involved in Stamens Development
The gene expression of TaGH9 in root, stem, leaf, spike and seed at the seedling, nutritional and reproductive stages were investigated using the most recent wheat transcriptome data. Available expression data are presented in Figure S4. The results show that most TaGH9s with stage-specific expression were differentially expressed in varied tissues. For example, TaGH9-16 is highly expressed in the root and stem.
Additional validation of the accuracy of TaGH9 in various tissues by qRT-PCR analysis was also performed on 24 TaGH9 genes to further characterize the accuracy of the TaGH9 in selected tissues. In accordance with transcriptome data, these TaGH9s differentially expressed in all five tissues (Figures 3 and S4). Interestingly, ten genes (TaGH9-2, TaGH9-4, TaGH9-7, TaGH9-9,  were highly expressed in the stamens (Figure 3).

TaGH9 Genes Were Involved in Stamens Development
The gene expression of TaGH9 in root, stem, leaf, spike and seed at the seedling, nutritional and reproductive stages were investigated using the most recent wheat transcriptome data. Available expression data are presented in Figure S4. The results show that most TaGH9s with stage-specific expression were differentially expressed in varied tissues. For example, TaGH9-16 is highly expressed in the root and stem.

TaGH9 Involving Pollen Development and Pollen Fertility Conversion
In our previous study, we found that TaGH9s were involved in pollen fertility conversion [31]. To investigate the roles of the TaGH9 family on anther and pollen development in BS366, pollen and anther morphology at critical stages of pollen development stage were observed. It was found that there was malformation when anther and pollen in BS366 were cultivated under sterile environment. As shown in Figure 4, there was no obvious difference between morphology of anther at critical stages of pollen growth (stages [6][7][8][9][10]. Under fertile condition, the anther could dehisce at stage 14 and be fertilized at stage 15 [32]. However, under sterile condition, anther indehiscence was showed at stage 14, then few or no pollen released from the anther, leading to unsuccessful fertilization in BS366 ( Figure 4). development in BS366, pollen and anther morphology at critical stages of pollen development stage were observed. It was found that there was malformation when anther and pollen in BS366 were cultivated under sterile environment. As shown in Figure 4, there was no obvious difference between morphology of anther at critical stages of pollen growth (stages 6-10). Under fertile condition, the anther could dehisce at stage 14 and be fertilized at stage 15 [32]. However, under sterile condition, anther indehiscence was showed at stage 14, then few or no pollen released from the anther, leading to unsuccessful fertilization in BS366 ( Figure 4).
To further understand the sterile characteristics of BS366, the pollen morphology under different fertility conditions was investigated. There was no difference between the microspore mother cell (stage 6) of fertile and sterile conditions. Compared with pollen under fertile condition, the cell plate at dyad (stage 7) under sterile condition cannot be properly formed in telophase I ( Figure 5A), resulting in an abnormal tetrad (stage 8). Moreover, we found no significant difference in stage 9. At the late uninucleate stage (stage 10), the starch content was decreased, and no generative or vegetative cells were observed. The expressions of these ten TaGH9s in different pollen development stages were also investigated. The results show that TaGH9-7, TaGH9-15 and TaGH9-19 genes highly expressed under the sterile environment during the critical stages of pollen fertility conversion ( Figure 5B).  To further understand the sterile characteristics of BS366, the pollen morphology under different fertility conditions was investigated. There was no difference between the microspore mother cell (stage 6) of fertile and sterile conditions. Compared with pollen under fertile condition, the cell plate at dyad (stage 7) under sterile condition cannot be properly formed in telophase I ( Figure 5A), resulting in an abnormal tetrad (stage 8). Moreover, we found no significant difference in stage 9. At the late uninucleate stage (stage 10), the starch content was decreased, and no generative or vegetative cells were observed. The expressions of these ten TaGH9s in different pollen development stages were also investigated. The results show that TaGH9-7, TaGH9-15 and TaGH9-19 genes highly expressed under the sterile environment during the critical stages of pollen fertility conversion ( Figure 5B). Micropores of sterile (a-e) and fertile BS366 pollen (f-j). Microspore mother cell (a,f), meiotic dyad (b,g), meiotic tetrad (c,h), early uninucleate stage (d,i) and late uninucleate stage (e,j). Bars = 20 μm. The data of expression profiles are from three biological replicates and three technical replicates, and error bars mean the standard error.

TaGH9 Involving Anther Dehiscence
Anther dehiscence is an important characteristic for male sterility of the BS366. To understand the roles of TaGH9, the expression of these genes was tested during anther dehiscence stages. It was found that TaGH9-2, TaGH9-4 and TaGH9-7 had lower expression under the sterile environment, while TaGH9-9 and TaGH9-15 had higher expression ( Figure 6A). Micropores of sterile (a-e) and fertile BS366 pollen (f-j). Microspore mother cell (a,f), meiotic dyad (b,g), meiotic tetrad (c,h), early uninucleate stage (d,i) and late uninucleate stage (e,j). Bars = 20 µm. The data of expression profiles are from three biological replicates and three technical replicates, and error bars mean the standard error.

TaGH9 Involving Anther Dehiscence
Anther dehiscence is an important characteristic for male sterility of the BS366. To understand the roles of TaGH9, the expression of these genes was tested during anther dehiscence stages. It was found that TaGH9-2, TaGH9-4 and TaGH9-7 had lower expression under the sterile environment, while TaGH9-9 and TaGH9-15 had higher expression ( Figure 6A). We also measured the cellulose content of anthers during the anthesis period, including stage 13-15 ( Figure 6B). The results show that the cellulose content of anthers was increasing during the anthesis period under fertile condition, while cellulose content was no longer increasing with flowering the four days before anthesis, suggesting that cellulose plays an important role in anther dehiscence. These results are highly consistent with the dynamic changes of TaGH9-2, TaGH9-4 and TaGH9-7 genes during the anther dehiscence period ( Figure 6A).
To characterize the anther dehiscence, anther wall was observed by SEM (scanning electron microscope) ( Figure 7). The results indicate that almost no cracking showed in anther, and the middle layer was smooth without wrinkles at the cracking of anther under sterile condition. However, anther could crack, and the middle layer had more wrinkles We also measured the cellulose content of anthers during the anthesis period, including stage 13-15 ( Figure 6B). The results show that the cellulose content of anthers was increasing during the anthesis period under fertile condition, while cellulose content was no longer increasing with flowering the four days before anthesis, suggesting that cellulose plays an important role in anther dehiscence. These results are highly consistent with the dynamic changes of TaGH9-2, TaGH9-4 and TaGH9-7 genes during the anther dehiscence period ( Figure 6A).
To characterize the anther dehiscence, anther wall was observed by SEM (scanning electron microscope) (Figure 7). The results indicate that almost no cracking showed in anther, and the middle layer was smooth without wrinkles at the cracking of anther under sterile condition. However, anther could crack, and the middle layer had more wrinkles at the cracking of anther under fertile condition. We also found there were more ubisch bodies in the inner anther wall under fertile condition than that under sterile condition. at the cracking of anther under fertile condition. We also found there were more ubisch bodies in the inner anther wall under fertile condition than that under sterile condition.

MicroRNA Targeting Prediction of TaGH9 and Expression Verification
miRNAs could interact with genes and play an essential function in plant growth and development [33]. In our earlier study, we found that miRNAs target genes, regulating male sterility in the PTGMS line BS366. To further understand the function of TaGH9 in the regulation of fertility, seven pollen and/or anther development-related TaGH9s were selected to evaluate the binding association with miRNAs. The results indicate that these seven genes harbored targets for ten different miRNAs (Figure 7 and Table S3). According to the discovery, we found that tae-miR5384-3p, tae-miR96574-5p, tae-miR9661-5p, tae-miR9676-5p and tae-miR398 potentially target TaGH9-15-A. It has been proved that miR398 could directly target Cu/Zn superoxide dismutase genes, which was associated with antioxidant stress [34]. In rice, miR398 may also interact with the expansion precursor protein gene (EXPB), involved in rice pollen germination [35]. In our previous study, the EXPB gene family could also interact with tae-aly-miRNA398a-5p/tae-blo-miRNA398a-5p, highly expressed in anthers under sterile environments, playing

MicroRNA Targeting Prediction of TaGH9 and Expression Verification
miRNAs could interact with genes and play an essential function in plant growth and development [33]. In our earlier study, we found that miRNAs target genes, regulating male sterility in the PTGMS line BS366. To further understand the function of TaGH9 in the regulation of fertility, seven pollen and/or anther development-related TaGH9s were selected to evaluate the binding association with miRNAs. The results indicate that these seven genes harbored targets for ten different miRNAs (Figure 7 and Table S3). According to the discovery, we found that tae-miR5384-3p, tae-miR96574-5p, tae-miR9661-5p, tae-miR9676-5p and tae-miR398 potentially target TaGH9-15-A. It has been proved that miR398 could directly target Cu/Zn superoxide dismutase genes, which was associated with antioxidant stress [34]. In rice, miR398 may also interact with the expansion precursor protein gene (EXPB), involved in rice pollen germination [35]. In our previous study, the EXPB gene family could also interact with tae-aly-miRNA398a-5p/tae-blo-miRNA398a-5p, highly expressed in anthers under sterile environments, playing important roles in fertility regulation [28]. In the present study, tea-miR398 also may target TaGH9-2-A/B/D. In addition, other known miRNAs such as tae-miR164 and tae-531 were predicted to target with TaGH9.

Discussion
Glycoside hydrolase family 9 (GH9) was previously reported as cellulase family E, whose members are spatiotemporal expression differences in plant growth and development and play various roles by encoding various hydrolysis-type genes and binding different substrates [5]. In recent years, GH9 have been identified from many plants such as rice (Oryza sativa L.), sorghum (Sorghum bicolor (L.) Moench), maize (Zea mays L.), shortstalked grass (Brachypodium distachyon (L.) Beauv.), Arabidopsis thaliana, poplar (Populus) and cotton (Gossypium. L). In our previous study, it was found that GH9 may be involved in male sterility by RNA-seq [31]. GH9 had differential alternative splicing between fertile and sterile condition, involved in fertility conversion. However, there are few reports regarding how GH9 works in pollen and anther development in crop plants. A total of 24 wheat GH9 with 74 copies were isolated in this study, to reveal the function of GH9 in wheat pollen and anther development (Figure 1 and Table S1).

TaGH9s Involving Phytohormone Crosstalk-Regulated Pollen Development
As molecular switches, cis-regulatory elements could respond to external stimuli, affecting plant growth by regulating gene transcription [36]. To further understand the functions of TaGH9, the cis-regulatory elements in promoter regions of TaGH9 were analyzed. It was found that four major cis-regulatory elements, including light-responsive elements, stress-related elements, growth and development-related elements and hormoneresponsive elements, were identified from TaGH9 ( Figure 2 and Table S2). Hormoneresponsive elements are enriched in these TaGH9s, especially MeJA response elements (TGACG-motif and CGTCA-motif) and abscisic acid response elements (ABRE element) are enriched in the TaGH9 family ( Figure 4). The latest research has informed us that GA and JA signaling are associated with anther dehiscence in BS366 [29,37]. Phytohormone crosstalk as a massive network has been known to be involved in plant growth. JA-GA crosstalk plays an essential role in stamen development and regulates male plant fertility [38,39]. Anther dehiscence involves cell abscission [40]. In our study, almost all TaGH9s have an ABRE element which is involved in the ABA signaling pathway, suggesting that ABA signaling is related to anther dehiscence besides JA signaling. JA and ABA signaling may have crosstalk in TaGH9, playing roles in pollen and anther development. The pollen of BS366 could sense light periods and low temperatures and cause fertility conversion [41]. We also found that nearly all TaGH9s contained G-boxes, which are involved in light signaling, suggesting TaGH9 may respond to light and be related to pollen development. To clarify the functions of TaGH9, expression profiles in different tissues ( Figure 3) were investigated. Ten highly expressed genes (including TaGH9-2, TaGH9-4, TaGH9-7, TaGH9-9, TaGH9-11, TaGH9-15, TaGH9-16, TaGH9-19, TaGH9-21 and TaGH9-22) in the stamens were identified based on the results of expression profiles of TaGH9s in tissues. These genes also contained more light and hormone-responsive elements.

TaGH9 May Be Involved in Anther Growth and Pollen Fertility Conversion
It is well known that the anther wall consists of five layers, including the epidermis, endothecium, middle layer (two layers) and tapetum [42]. Lipid and polysaccharide metabolism are required for development of the anther wall, since polysaccharide metabolism could stimulate formation and degradation of different anther wall layers [43]. The pollen tube wall is formed by the extension of the anther inner wall, which is mainly composed of pectin polymers, cellulose and hemicellulose, involved in polysaccharide metabolism [44]. The deposition of cellulose could be catalyzed by the cellulose synthase complex (CSC), which consists of multiple individual cellulose synthase (CESA) proteins, synthesizing at the plasma membrane via the cellulose synthase (CESA) complex [45]. It has been reported that the catalytic core of CSC consists of three types of CESA subunits, including CESA1, CESA3, and CESA6, as a deficiency in either component results in a reduction in cellulose and abnormal amounts of pollen, leading to male sterility, suggesting that cellulose synthesis is necessary for pollen development [46]. In rice, the mutation of Os12BGlu38, a member of GH1, showed shrunken and hollow pollen and deficient mutant pollen. In addition, no inner wall was observed in the pollen of Os12BGlu38 mutants using transmission electron microscopy (TEM) [47]. Several GH9 have been charactered in Arabidopsis and rice. OsGH9B18, OsGH9B6/14 and OsGH9C2 in rice, At3g43860, AtGH9A3, AtGH9A4 and AtGH9B5/11/14/17 in Arabidopsis, specifically expressed in stamen and pollen [8,48]. Genes with close homology may have similar roles. In our study, we found that TaGH9-4/7/9 classified into Class C with OsGH9C2 and TaGH9-2/19/15/11 classified into Class B1 with AtGH9B14, AtGH9A3 and OsGH9B18, suggesting that these TaGH9 genes may have a similar function with genes in rice and Arabidopsis (Figure 1). TaGH9-19 especially has very close homology with AtGH9B14, AtGH9A3 and OsGH9B18. We also found that miR164 may target TaGH9-9 genes (Figure 8). In Arabidopsis, miR164 cleaves ORE1, which is a transcription factor of NO APICAL MERISTEM (NAM), participating in age-related cell death [49]. A lack of miR164-CUC1/2 results in the defect in sepals and low fertility in plants [50,51]. In addition, miR164 is induced by UV-B radiation in maize [52]. In our previous study, we also found miR164 was down-regulated in the pollen at the meiosis stage in sterile BS366 [28]. These results suggest that TaGH9-9 genes interacted with miR164, participating in pollen fertility conversion by responding to light ( Figure 8).
LSSR1, encoding a GH5 cellulase, expressed in the anther microsporogenesis stage of rice. It has been reported that LSSR1 with an N terminal signal peptide is secretory protein stored in pollen grains and plays important roles during fertilization in rice. The results indicate that the process of fertilization was inhibited in the mutant LSSR1, resulting in reduced fertility in rice, indicating that LSSR1 plays an essential role in male fertility in rice [53]. In our previous study, following the formation of the cell plate, pectin and xyloglucan gradually combined, excess membrane began to disappear, cellulose content increased and eventually cell walls formed [54][55][56]. During meiotic cytokinesis, cellulose accumulates at both the division place and external wall of the cell [57]. At the prophase I stage, callose deposition happened in the pollen mother cell (PMC), forming a thick callose cell wall until the end of meiosis cell cycle [58]. Cellulose deposition is required for spore development during meiosis since it provides the basis for later pollen cell wall formation [56]. It has been reported that lack of cellulose deposition during meiosis could cause male sterility [59]. In our study, the cell plate abnormally formed in stage 7 (dyad) and stage 8 (tetrad) (Figure 5A), and homologous chromosomes failing to pair together were found in BS366 under sterile condition. In addition, TaGH9-7, TaGH9-15 and TaGH9-19 were highly expressed in the pollen development stage, especially in stage 7, 8 and 10, which were the critical stages of fertility conversion, indicating that highly expressed TaGH9-7, TaGH9-15 and TaGH9-19 during pollen development may be one of the reasons for pollen abortion ( Figure 5A,B).

TaGH9 May Be Involved in Anther Dehiscence
Anther dehiscence depends on the following processes: expansion of the endothelium, deposition of fibrous bands in the inner layer, degradation of the septal cells and cracking of the dehiscence [40]. It has been reported that the content of cellulase in anther increases significantly during the development of anthers and pollens. Cellulase may also be involved in wall morphogenesis such as tapetum wall and callose enzyme activity. The timing of the callose enzyme activity is critical for proper male gametophyte development and the wrong timing may lead to male sterility [60]. In Arabidopsis, the KORRIGAN allele of GH9A-rsw2-1 mutants showed that reductions in cellulose production, constitutively slow root growth and enhanced temperature-sensitive responses shortened stamen filaments and impaired anther dehiscence, while all characteristics were restored when the 8.5kb mucilage containing the KORRIGAN gene was transformed into the mutant rsw2-1 [61]. In addition, overexpression of PtrCel9A6, a member of GH9, causes an increase in glycoside hydrolase activity, interferes with the accumulation of cellulose in epidermal cells and inhibita the thickening of the intra-anther wall, preventing normal anther dehiscence and pollen release, resulting in pollen grains remaining in their pollen sacs and ultimately leading to reduced fertility [62]. Moreover, the dehydration of anther wall cells also contributes to anther dehiscence. The cells of the anther wall dehydrate due to the decrease in water potential in the loculus. During the process of anther dehiscence, water content in anther cells decreased by transpiration, meanwhile, the transformation of starch into sugar increased the osmotic potential of anther tissue, promoting the success of the anther dehiscence process [63]. In tobacco, the defect of PIP2-like water channel proteins causes blocking of the dehydration process, resulting in abnormal anthers dehiscence [64]. Furthermore, in Arabidopsis, the H + -sucrose transporter AtSUC1, observed in the anther septum, could increase the osmotic potential of the septum and cause dehydration in the anther [65]. It was also reported that the concentration levels of flavonoid transport factor FFT in anther are involved in anther dehydration in Arabidopsis [66]. In our study, compared with anther dehiscence under fertile condition, the middle layer of anther was smooth, and few wrinkles were observed in indehiscent anther under sterile condition (Figure 7). The content of cellulose in anther under sterile condition was not increase with flowering, while it increased under fertile condition. In addition, highly expressed TaGH9-9 and TaGH9-15 at the anther dehiscence stage under sterile condition increased the osmotic potential of anthers wall cells, causing anther to fail to crack. Meanwhile, the down regulation of TaGH9-4 and TaGH9-7 may reduce the degradation rate of cellulose and failure to degrade septum, resulting in anther indehiscence.
In the myb26 mutant of Arabidopsis, there is no radial expansion and secondary thickening of the wall cells within the anther chamber, resulting in the inability of the anther dehiscence, causing male sterility [67]. In wheat, it has been reported that miR398 targets the EXPB gene, up-regulating sterile anther involving pollen fertility conversion and pollen germination ability [28,68]. MiR398 could negatively interact with TaGH9-2 and TaGH9-15 at the anther dehiscence stage, indicating that these two genes may regulate the expansion of the cell wall and be involved in anther dehiscence (Figure 8 C).

A Putative Pathway of TaGH9 Regulating Pollen Fertility Conversion and Anther Dehiscence in Wheat
In this research, we formulated a possible regulatory pathway of TaGH9 in pollen development and anther dehiscence (Figure 9). After stimulation by external environment with short day and low temperature at meiosis stage, wheat PTGMS line BS366 shows abnormal growth in pollen and anther. For pollen development, tae-miR164, tae-miR9661-5p, tae-miR5384-3p and tae-miR9676-5p showed competitive expression with target TaGH9s under sterile conditions. These TaGH9s were also involved in phytohormone signaling, such as JA-ABA crosstalk, contributing to the meiosis abnormalities and resulting in male sterility. Moreover, TaGH9 is associated with anther dehiscence. In this study, we found that tae-miR164, tae-miR398 and tae-miR9661-5p targeted TaGH9-9, TaGH9-15, TaGH9-2 and TaGH9-4, respectively, involving the cellulose content of the anther wall. The reduction in cellulose content in the anther wall resulted in anther indehiscence. Collectively, these findings suggested that TaGH9s might have a significant effect in male sterility in wheat PTGMS line BS366.
Samples for fertility and anther dehiscence analysis were collected using the following equation reported by [26,37].
For tissue-specific expression analysis, the root, stem, leaf, petal, pistil, stamen and glume of wheat at the meiosis stage were collected.

Phenotypic Characterization of BS366
Anthers at different stages from fertile and sterile conditions were photographed with a ZEISS SteREO Discovery.V20 dissecting microscope (ZEISS, Jena, Germany). To further analyze pollen fertility conversion, the anthers of BS366 from meiosis to the mature pollen stage under both fertile and sterile conditions were collected and fixed in FAA solution (formaldehyde: glacial acetic acid: 50% ethanol = 5: 5: 9). The pollen grains were released from anthers by tweezers and dyed with improved carbol fuchsin solution [62] Photographs of microspores and pollen were obtained using an LEICA DM6000B microscope (LEICA, Wetzlar, Germany). For scanning electron microscope (SEM) analysis, anthers at the dehiscence stage were collected, fixed in 2.5% glutaraldehyde, Figure 9. A putative pathway of TaGH9 regulating to pollen fertility conversion and anther dehiscence in wheat. Organelles under orange ovals with gene names correspond to predicted subcellular locations for TaGH9.

Plant Material, Growing Conditions and Sample Collection
In this study, the wheat photo-thermo-sensitive genetic male sterile (PTGMS) line BS366 (winter wheat) was used. Growing conditions: The seeds were planted in October 2020 in the experimental fields in Beijing (N 39 • 54 , E 116 • 18 ) and Dengzhou (N 32 • 22 , E 111 • 37 ), China, respectively, and managed conventionally.
Samples for fertility and anther dehiscence analysis were collected using the following equation reported by [26,37].
For tissue-specific expression analysis, the root, stem, leaf, petal, pistil, stamen and glume of wheat at the meiosis stage were collected.

Phenotypic Characterization of BS366
Anthers at different stages from fertile and sterile conditions were photographed with a ZEISS SteREO Discovery.V20 dissecting microscope (ZEISS, Jena, Germany). To further analyze pollen fertility conversion, the anthers of BS366 from meiosis to the mature pollen stage under both fertile and sterile conditions were collected and fixed in FAA solution (formaldehyde: glacial acetic acid: 50% ethanol = 5: 5: 9). The pollen grains were released from anthers by tweezers and dyed with improved carbol fuchsin solution [62] Photographs of microspores and pollen were obtained using an LEICA DM6000B microscope (LEICA, Wetzlar, Germany). For scanning electron microscope (SEM) analysis, anthers at the dehiscence stage were collected, fixed in 2.5% glutaraldehyde, dehydrated, air dried in silica, coated with gold-platinum in a sputter coater and finally examined by SEM (Phenom LE, Wetzlar, Germany) [69].

Identification of Wheat GH9 Family Members
The Ex PASy-ProtParam online tool (https://web.expasy.org/protparam/, accessed on 13 February 2022) was used to analyze the basic physicochemical properties of the final obtained wheat GH9 protein, including gene CDS, amino acid number (AA), relative molecular weight (MW), isoelectric point (PI), total mean hydrophilicity value (GRAVY) and aliphatic index (AI). The subcellular localization of the wheat GH9 protein was predicted using the target P 1.1 Server (http://www.cbs.dtu.dk/services/Target~P/, accessed on 13 February 2022). GH9 subgenomic copies were named A, B and D according to their position on the chromosome.

Chromosomal Locations and Synteny Analysis
Positional information of TaGH9 genes were collected based on the genome annotation information. The synteny analysis of TaGH9 genes was performed by using the whole genome synteny block data, which was available in the Plant Genome Duplication Database (http://chibba.agtec.uga.edu/, accessed on 20 February 2022) [70]. The visualization of chromosomal locations of the TaGH9 family was carried out by using Circos-0.69 (http: //circos.ca/, accessed on 20 February 2022) [71].

Multiple Sequence Alignment and Phylogenetic Tree Construction
Multiple alignment of GH9 protein sequences was performed by using DNAMAN (ver. 6.0, LynnonBiosoft, CA, USA) with default parameters. The phylogenetic tree was constructed by using MEGA6.0 [49] with the neighbor-joining (NJ) method (1000 bootstrap trials and the Poisson model) according to the results of multiple alignment.

Structural Analysis of TaGH9 Genes and Proteins
Gene structure information of the TaGH9 family members were visualized using the Gene Structure Display Server (GSDS) tool [72] (http://gsds.cbi.pku.edu.cn/, accessed on 25 February 2022) by comparing the predicted GH9 coding sequences with their corresponding genomic sequences.

Analysis of cis-Regulatory Elements of GH9 Genes
To identify the cis-elements in the promoter sequences of the GH9 family genes in wheat, 1.5-kb of genomic sequences upstream of the promoter regions of each GH9 genes was chosen and screened against the Plant CARE database [73] (http://bioinformatics.psb.ugent.be/ webtools/plantcare/html/, accessed on 25 February 2022) to identify the cis-elements.

RNA-Seq Data Analysis and Gene Expression Heatmap
The RNA-seq data of six tissues (root, stem, leaf, pistil, stamen and glume) and three developmental periods (seedling, vegetative and reproductive) of Chinese spring wheat were downloaded and analyzed using expVIP (http://wheat-expression.com, accessed on 27 February 2022). The TPM (Transcripts Per Kilobase of exon model per million mapped reads) data of GH9 genes were visualized with a heat map using Graphpad Prism 9 software.

Measurement of Cellulose Content in Anther
Fresh wheat anthers of stage 13, 14 and 15 were collected and ground into powder under liquid nitrogen. The anthrone reagent was used to determine the cellulose content after constructing a standard curve for the quantification of cellulose [74].
4.11. miRNA Targeting Prediction of the TaGH9 Family miRNA could play an important role in plant development by targeting genes. To investigate whether the candidate TaGH9 interacts with miRNA, the known and studied miRNA sequences of wheat were collected from miRbase and subjected to online psR-NATarget. The interaction networks between the candidate TaGH9 and miRNAs were visualized by using Cytoscape (version 3.7, CA, USA) based on predicted results.

Expression Analysis of TaGH9 and miRNAs
The total RNA of all samples was extracted using TRIzol reagent (Invitrogen, CA, USA) following the manufactures' instructions, and approximately 1 µg RNA was reverse transcribed with PrimerScript 1st Strand cDNA synthesis kit (TaKaRa, Shiga, Japan) and miRcute Plus miRNA cDNA T (Tingen, Beijing, China) for genes and miRNAs, respectively, according to the manufacturer protocol. qRT-RCR assays for TaGH9 and miRNAs were conducted on an Eco Real-time PCR system (Illumina, CA, USA) using TB Green Premix Ex Taq (Tli RNaseH Plus) kit (TaKaRa, Shiga, Japan) and SYBR ® miRcute Plus miRNA PreMix (Tingen, Beijing, China), respectively, as described previously [28,75]. Wheat 18S gene served as the reference gene for TaGH9 analysis, and U6 was used as a reference gene for miRNA analysis. The relative expression levels were calculated according to the 2 -∆∆Ct method for relative expression with three biological replicates and three technical replicates [76]. The primers used in this study were designed using the Primer premier 5.0 (Primer, CAN, UK) program and are listed in Table S4.

Conclusions
In this study, a comprehensive overview of the TaGH9 gene family in wheat, including gene and protein structures, phylogenetic relationships and expression profiles, was provided. A total 74 copies of TaGH9 were identified and classified into three major subclades, named Class A, Class B and Class C. The TaGH9s in the same subfamily and subclades may own similar functions. There were abundant light-response-related elements and hormone-response-related elements in the promoter region that assist TaGH9s to respond to light and hormones, regulating anther and pollen development. Ten TaGH9s were highly expressed in stamens. Finally, seven TaGH9s may be associated with pollen development and anther dehiscence. A putative pathway of TaGH9s with targeted miRNA regulating to pollen fertility conversion and anther dehiscence was constructed. The potential function of TaGH9 needs to be further verified using more molecular biological techniques and tools. Collectively, these findings establish a foundation for further exploration of TaGH9 and provide novel insights into its biological functions.