Mutation of glucose-methanol-choline oxidoreductase leads to thermosensitive genic male sterility in rice and Arabidopsis.

Thermosensitive genic male sterility (TGMS) lines serve as the major genetic resource for two-line hybrid breeding in rice. However, their unstable sterility under occasional low temperatures in summer highly limits their application. In this study, we identified a novel rice TGMS line, ostms18, of cultivar ZH11 (Oryza sativa ssp. japonica). ostms18 sterility is more stable in summer than the TGMS line carrying the widely used locus tms5 in the ZH11 genetic background, suggesting its potential application for rice breeding. The ostms18 TGMS trait is caused by the point mutation from Gly to Ser in a glucose-methanol-choline (GMC) oxidoreductase; knockout of the oxidoreductase was previously reported to cause complete male sterility. Cellular analysis revealed the pollen wall of ostms18 to be defective, leading to aborted pollen under high temperature. Further analysis showed that the tapetal transcription factor OsMS188 directly regulates OsTMS18 for pollen wall formation. Under low temperature, the flawed pollen wall in ostms18 is sufficient to protect its microspore, allowing for development of functional pollen and restoring fertility. We identified the orthologous gene in Arabidopsis. Although mutants for the gene were fertile under normal conditions (24°C), fertility was significantly reduced under high temperature (28°C), exhibiting a TGMS trait. A cellular mechanism integrated with genetic mutations and different plant species for fertility restoration of TGMS lines is proposed.

Introduction

Male‐sterile lines in plants are useful for hybrid breeding to promote heterosis in crop production and protect the intellectual property of germplasm in commercial seed production (Ma, 2005). Male sterility consists of cytoplasmic male sterility (CMS) and genic male sterility (GMS). CMS lines facilitated the establishment of a three‐line breeding system in the 1970s, which is still widely applied in rice hybrid production (Chen and Liu, 2014; Cheng et al., 2007). In 1973, a Chinese researcher accidentally discovered the photoperiod‐sensitive male‐sterile (PGMS) line Nongken 58S (NK58S) of Japonica rice (Oryza sativa ssp. japonica). After introgressive backcrossing of NK58S with the indica background, the product Peiai 64S (PA64S) displayed a temperature‐sensitive male‐sterile (TGMS) trait (Yang et al., 2009). Subsequently, Deng et al. identified the spontaneous mutant AnnongS‐1 (AnS‐1) of Indica rice (Oryza sativa ssp. indica) as a TGMS line (Deng et al., 1999). In general, photoperiod/thermosensitive GMS (P/TGMS) lines show male sterility under restrictive conditions (high temperature or long‐day photoperiod) but male fertility under permissive conditions (low temperature or short‐day photoperiod). Based on P/TGMS plants, a two‐line system has been developed for rice breeding; P/TGMS plants can receive restorers with a wider range of germplasm resources compared with the CMS. It is estimated that the average yield of two‐line hybrid rice is 5%–10% higher than that of three‐line hybrid rice (Yang et al., 2009; Yuan, 2004). Overall, the two‐line hybrid system is an important innovation and is anticipated to be a more popular method for hybrid seed production in China (Huang et al., 2014; Si et al., 2011; Yuan, 1990); indeed, the growing area of the two‐line system has reached half of the total hybrid rice growing area (China National Rice Research Institute, 2019).

Several P/TGMS genes have been identified in recent years. In NK58S and Peiai 64S, a substitution of C‐to‐G in the PMS3/P/TMS12‐1 locus encoding a 21‐nucleotide small RNA leads to the male‐sterile phenotype under a long‐day photoperiod and high temperature, respectively (Ding et al., 2012; Zhou et al., 2012). The TGMS trait of AnS‐1 is caused by C‐to‐A mutation in TMS5, which causes a premature stop codon in the RNase Z protein (Zhou et al., 2014). RNA interference or cosuppression of the UDP‐glucose pyrophosphorylase1 (UGP1) results in a thermosensitive genic male sterility (Chen et al., 2007); a frame shift in a rice leucine‐rich repeat–receptor‐like kinase, TMS10, also leads to thermosensitive male sterility (Yu et al., 2017). Although several P/TGMS genes have been cloned, the common mechanism that explains how temperature or photoperiod affects the fertility of P/TGMS lines in rice remains unclear. After decades of efforts, many TGMS lines have been used in two‐line rice breeding. Among them, tms5 is currently a predominant TGMS locus for rice two‐line breeding. Hybrid rice cultivars containing the tms5 locus account for at least 71% of two‐line rice cultivars and 83.8% of the land‐growing two‐line hybrid rice in China (Zhou et al., 2014). In rice cultivation, TGMS lines are grown in the field in summer, with high temperatures; they exhibit a sterile phenotype and are used for crosses with other rice varieties to produce hybrid seeds. With global climate changes, severe temperature changes frequently occur in summer (Hasegawa et al., 2021), and occasional low‐temperature events in summer lead to incomplete male sterility, severely reducing the purity of hybrid seeds (Si et al., 2011). TGMS lines with more stable sterility are in high demand for the application of two‐line systems for hybrid seed production.

In this study, we identified a novel rice TGMS line, ostms18, which was obtained by EMS mutagenesis screening. Field experiment results demonstrate that ostms18 exhibits more stable sterility in summer than tms5, with great value for hybrid seed production. Molecular cloning revealed that a point mutation in a gene encoding glucose‐methanol‐choline (GMC) oxidoreductase led to the TGMS phenotype. This gene is required for pollen wall formation, which is directly regulated by OsMS188, a transcription factor in the tapetum. The mutant of AtTMS18, a homologous gene in Arabidopsis, also exhibits a TGMS phenotype. Based on these results, we provide a cellular mechanism for fertility restoration that integrates genetic mutations and plant species.

Results

The ostms18 locus provides a novel genetic resource for hybrid rice breeding

We conducted EMS mutagenesis of ZH11 (Oryza sativa ssp. japonica) and isolated several TGMS lines. One of them, named ostms18, exhibited complete male sterility under high temperature (over 29°C) but fertility under low temperature (below 23°C) (Figure 1b,c). There was no difference between wildtype and mutant during the vegetative growth stage (Figure 1a–c). At the flowering stage, the ostms18 anthers were shrunken under high temperature; at low temperature, the anthers were plump and yellow in restored plants (Figure 1d,e). Alexander staining showed that no mature pollen was presented in the shrunken anthers but that large numbers of mature pollen were present in the plump anthers (Figure 1f).

Figure 1

The TGMS traits of ostms18. (a–c) The wild‐type (WT) plant and ostms18 mutant after the heading stage under different temperatures. HT, High temperature (over 29°C); LT, Low temperature (below 23°C). Bar = 10 cm. (d) The spikelets of the WT plant and ostms18 mutant after removing the palea. Bar = 1 mm. (e) The anther of WT plant and ostms18 mutant. (f) Alexander staining of the WT plant and ostms18 anther. Bar = 200 μm. (g) The panicles of ostms18 and tms5 under high temperature season in 2021. Bar = 2 cm. (h–i) The seed setting rates of different batches of ostms18 and tms5 (ZH11) booting under high temperature season (from 7/30 to 8/10) (h) and low temperature season (from 9/1 to 9/13) (i) in 2021. The numbers on the abscissa represent the booting dates. AT, Ambient temperature. [Colour figure can be viewed at wileyonlinelibrary.com]

As TGMS lines carrying the locus tms5 are widely used in two‐line hybrid breeding (Zhou et al., 2014), we compared TGMS traits between ostms18 and AnS‐1 (Oryza sativa ssp. indica). The seeds of ostms18 and AnS‐1 germinated once a week and were transplanted into the paddy field 3 weeks later. Under the high‐temperature season in summer 2018 (from the end of July to the middle of August) (over 29°C), both ostms18 and AnS‐1 displayed very low fertility, whereas the seed‐setting rate of ostms18 was lower than that of AnS‐1 (Figure S1a, batches 1–7). This result indicated that ostms18 showed more stable male sterility than AnS‐1, even though they are in different genetic backgrounds (Figure S1a). In addition, the seed‐setting rate began to recover gradually as the average temperature decreased (below 26°C, the 8th batch) (Figure S1a). These data suggest that the efficiency of fertility recovery of ostms18 is similar to that of AnS‐1. To reduce the influence of genetic background on fertility recovery, we further compared the fertility of ostms18 with TMS5 knockout line in the ZH11 background. Under high temperature in 2020 (over 29°C), the sterility rate of ostms18 was similar to that of tms5 (ZH11) (Figure S1b). During the high‐temperature season in 2021, the continuous unusual weather that persisted in Shanghai caused an occasional low‐temperature season (the average temperature was below 28.2°C), which led to a relatively higher selfing seeds rate in both tms5 (ZH11) and ostms18. The selfing rate of tms5 (ZH11) was 13.33%–15.03% and that of ostms18 was 2.26%–5.85%, which was much smaller than that of tms5 (Figure 1g,h). The above results suggest that the sterility of ostms18 is more stable than that of tms5 (ZH11). Nevertheless, the effect of low temperature on the fertility recovery of ostms18 was similar to that of tms5 (Figure 1i and S1b). The more stable sterility of ostms18 in summer indicates the high application value of this genetic locus for rice two‐line breeding.

encodes a glucose‐methanol‐choline (GMC) oxidoreductase

ostms18 was crossed with the wild type, and the F1 plants exhibited normal fertility. Fertile and sterile plants in the F2 population segregated at a 405:110 ratio (χ2 = 3.78 for 3:1, P > 0.05), indicating that the inherited male sterile phenotype is a single recessive Mendelian locus. To identify a candidate gene, we generated a population by crossing ostms18 (ZH11) and wildtype Kasalath (an Indica cultivar). For first‐pass mapping, F2 male‐sterile lines were divided into four DNA pools combined with the parent DNA control. A total of 80 insertion/deletion (In/Del) markers were used (average 7 markers per chromosome) (Wang et al., 2011), and the OsTMS18 gene was mapped to a 590‐kb region between In/Del markers 20.11 M and 20.70 M on chromosome 10 (Figure S2). Another population was generated from a cross between ostms18 and wildtype ZH11 for bulked segregant analysis (BSA). The results revealed a point mutation, GGC (Gly) to AGC (Ser), in the second exon of LOC_Os10g38050 (Figure 2a,b). We then introduced the genomic DNA fragment of LOC_Os10g38050 with its predicted promoter into the ostms18 mutant to carry out a complementation experiment. Pollen formation was restored in all 32 independent transgenic plants under high temperature (Figure 2f–h), confirming that the mutation within LOC_Os10g38050 is responsible for the ostms18 phenotype.

Figure 2

OsTMS18 encodes an anther‐expressed GMC oxidoreductase. (a) SNP index for mapping the OsTMS18 by the BSA‐seq approach. (b) Identification of the mutated point of OsTMS18 in a WT plant and ostms18 mutant by sequencing (position 181). SEM observation for the pollen grains from dehiscence anther of ostms18 mutant (c–e) and the transgenic plants for complementation of ostms18 (f–h). Bars = 100 μm in (c, f). Bars = 5 μm in (d, g). Bars = 1 μm in (e, h) (i–n) Confocal images of the fluorescence of the OsTMS18‐GFP fusion proteins from stages 8 through 12. GFP expression (530 nm) is shown in the green channel, while chlorophyll autofluorescence (>560 nm) is shown in the red channel. Bars = 20 μm in (i–m). Bar = 2 μm in (n). [Colour figure can be viewed at wileyonlinelibrary.com]

OsTMS18 encodes a glucose‐methanol‐choline (GMC) oxidoreductase that was previously identified as NP1. All np1 alleles, including np1‐1 ~ np1‐4, result in a complete male‐sterile phenotype. In situ hybridization showed transcripts expression in the tapetum and microspores (Chang et al., 2016; Liu et al., 2017). To further understand its expression, we fused the genomic DNA sequences of OsTMS18 to GFP driven by its native promoter. This construct also complemented the male‐sterile phenotype of ostms18 under high temperature. In this transgenic line, the OsTMS18‐GFP signal was initially detected in the tapetum and the surface of microspores at stage 9 (Figure 2j). At late stages, OsTMS18‐GFP fluorescence was significantly increased in the tapetum and locule (Figure 2k–m). Finally, the OsTMS18‐GFP fusion protein was accumulated highly on the surface of developing microspores (Figure 2n), though the OsTMS18‐GFP signal was not detected inside the microspores. Overall, expression of OsTMS18 in anthers is consistent with a role in pollen formation.

Low temperature overcomes pollen wall defects to produce functional pollen in ostms18

With np1 alleles, mature pollen grains are rarely detected in their small and whitish anthers of these plants (Chang et al., 2016; Liu et al., 2017). Under high temperature, the shrivelled anther of ostms18 contained collapsed pollen remnants (Figure S3). SEM showed that the exine layer of ostms18 pollen remnants has an obviously cracked appearance (Figure 3c,d). In restored ostms18 plant grown under low temperature, pollen exhibits intact an exine layer similar to that of wildtype (Figure 3e,f). In rice, anther development is divided into 14 stages (Zhang and Wilson, 2009). Microspores are released from the tetrad at stage 9, become vacuolized at stage 10, and form a crescent structure at stage 11 (Zhang and Wilson, 2009). Before stage 10, no detectable difference was observed between ostms18 and wildtype (Figure 3j and S4), consistent with previous cellular observations for np1 alleles (Chang et al., 2016; Liu et al., 2017). Under high temperature, ostms18 microspores still exhibited abnormal vacuolization and failed to form the crescent structure at stage 11 (Figure 3k). Finally, the microspores ruptured at stage 13 (Figure 3l). Under low temperature, microspore vacuolization and crescent structure were normal in the ostms18 mutant, and pollen grains mostly formed eventually (Figure 3m–o). These cytological results indicate that the microspore development defect in ostms18 under high temperature is overcome under low‐temperature conditions.

Figure 3

Cytological analysis of anther development in the WT and ostms18. (a, c, e) SEM observation for single pollen of WT (a) and ostms18 at high temperature (c), and at low temperature (e). (b, d, f) The enlarged view of the surface of pollen grains of WT (b) and ostms18 at high temperature (d), and at low temperature (f). HT, High temperature (>29°C); LT, Low temperature (<23°C). Bars = 10 μm in (a, c and e), 1 μm in (b, d and f). (g–i) Semi‐thin cross‐sectional analysis of WT anther during the anther development stages. (j–l) Semi‐thin cross‐sectional analysis of ostms18 mutant anther at high temperature (>29°C) during the anther development stages. (m‐o) Semi‐thin cross‐sectional analysis of ostms18 mutant anther at low temperature (<23°C) during the anther development stages. BMs, bicellular microspore; Dp, degenerated pollen; HT, High temperature (>29°C); LT, Low temperature (<23°C); E, epidermis; En, endothecium; Ms, microspore; Mp, mature pollen; T, tapetum. Bars = 20 μm. (p–r) TEM observation for WT pollen development from stages 9, 10, 11. (s–u) TEM observation for ostms18 pollen development at high temperature from stages 9, 10, 11. (v–x) TEM observation for ostms18 pollen development at low temperature from stages 9, 10, 11. The boxed image on the right of each panel was enlarged from the left region. Ba, bacula; Ex, exine; Msp, microspore; Ne, nexine; Se, sexine. Bars = 5 μm in the left of panel and 1 μm in the right of panel (p–x). [Colour figure can be viewed at wileyonlinelibrary.com]

Transmission electron microscopy (TEM) analysis was further performed to examine the detailed defects of ostms18. After microspores are released from the tetrad at stage 9, electron‐dense probacular materials were regularly deposited outside the plasma membrane in wildtype (Figure 3p). Under high temperature, electron‐dense granules accumulated in a disorderly manner on the surface of ostms18 microspores (Figure 3s). At stage 10, the exine was formed in wildtype, with a distinctive two‐layer structure (Figure 3q). Although this structure was present on ostms18 microspores, its inner layer was obviously much thinner than the outer layer (Figure 3t). At stage 11, the wildtype microspore exhibited a crescent shape with a well‐organized exine (Figure 3r). Conversely, the ostms18 microspore collapsed, and its exine was irregularly thickened (Figure 3u). Therefore, the irregular structure of the exine is the cause of microspore leakage and abortion. Under low temperature, the exine deposition on the surface of ostms18 microspores was normal at stage 9 (Figure 3v). Subsequently, the two‐layer structure could be formed in the mutant, though they were thinner than those of wildtype (Figure 3w). At stage 11, the microspore with a thinner two‐layer structure formed a crescent shape, similar to wildtype (Figure 3x). This result suggests that the defective exine structure in ostms18 leads to pollen abortion under high temperature. Under low temperature, the two‐layer structure of exine eventually formed, which supports the microspores of ostms18 developing into functional pollen.

Mutation of the homologue in Arabidopsis leads to the TGMS phenotype

OsTMS18 shows high sequence similarity with three genes (AT1G12570, AT1G73050, and AT3G56060) in Arabidopsis. We produced constructs containing these genes fused with GFP driven by their native promoters and transformed them into wildtype Arabidopsis Col plants. AT1G12570‐GFP was detected in the tapetum and anther locule after microspore release from the tetrad, and fluorescence gradually accumulated on the surface of the microspore at anther late stages (Figure 4a). This expression profile is quite similar to that of OsTMS18 in rice. AT1G73050 and AT3G56060 were expressed in microspore mother cells and tetrads, which is quite different from OsTMS18 expression in the tapetum (Figure S5). Therefore, we designated the AT1G12570 gene AtTMS18. The knockout line of AtTMS18 was obtained through the CRISPR‐Cas9 approach. This line carries a 7‐bp deletion (attms18–1) in the second exon, which leads to a frameshift in the encoding region (Figure S6). Another allele (attms18–2) was obtained from the Salk mutant library in which the T‐DNA is inserted in the fourth intron of AtTMS18 (Figure S6). Under normal temperature (24°C), these two lines were fertile, similar to wildtype (Figure 4c,d); under high temperature (28°C), fertilities were only 30% of that of wildtype (Figure 4e,f). SEM showed collapse of nearly half of the pollen grains, with an abnormal sculpted structure (Figure S7). These results reveal that the mutation of OsTMS18 homologues leads to the TGMS phenotype in Arabidopsis.

Figure 4

The attms18 also exhibits the TGMS traits. (a, b) Confocal images of the fluorescence of the AtTMS18‐GFP fusion proteins in wild type (a) and ms188 (b) from stages 7 through 11. GFP expression (530 nm) is shown in the green channel, while chlorophyll autofluorescence (>560 nm) is shown in the red channel. Bars = 20 μm. (c) Fertile phenotypes of WT, attms18‐1 and attms18‐2 grown at at LT (24°C). (d) The fertility ratio of wild type, attms18‐1 and attms18‐2 at LT (24°C). (e) Fertile phenotypes of WT, attms18‐1 and attms18‐2 grown at HT (28°C). (f) The fertility ratio of wild type, attms18‐1 and attms18‐2 at HT (28°C). Insets in (c, e) show the viable pollen of anthers stained purple by Alexander's staining. S, sterility; PS, partial sterility; F, Fertility. Bars = 2 cm. [Colour figure can be viewed at wileyonlinelibrary.com]

Expression of is directly regulated by

Several transcription factors, including UDT1, OsTDF1, TDR, bHLH142, EAT1 and OsMS188, have been reported to be essential for rice tapetum development. These genes form a regulatory pathway with OsMS188 located downstream of the pathway (Cai et al., 2015; Han et al., 2021; Jung et al., 2005; Ko et al., 2014; Li et al., 2006; Niu et al., 2013). Their orthologs in Arabidopsis form a genetic pathway for tapetal development and pollen wall formation (Gu et al., 2014; Lou et al., 2014; Lou et al., 2018; Zhu et al., 2011). We analysed the relationship between OsTMS18 and these transcription factors. Both RT‐PCR and qRT‐PCR analyses showed OsTMS18 expression to be significantly down‐regulated in all of these mutants, including osms188 (Figure S8). In Arabidopsis, AtTMS18 expression was also significantly suppressed in the ms188 mutant (Figure 4b). These data show that both OsTMS18 and AtTMS18 act downstream of OsMS188 and MS188 in rice and Arabidopsis, respectively. The promoter regions of OsTMS18 and AtTMS18 contain the core motifs of MYB (AACC) cis‐elements (Figure 5a,b). OsMS188 and MS188 proteins were expressed and purified from Rosetta Escherichia coli, and electrophoretic mobility shift assays (EMSAs) showed that the OsMS188 and MS188 proteins bind to DNA probes containing the ‘AACC’ core motifs. Moreover, DNA binding was competed by 20‐ and 200‐fold unlabeled probe (Figure 5a,b), but the abundance of the shifted band was not reduced by unlabeled competitor (Figure 5a,b). A transient expression assay using Arabidopsis protoplasts was performed to analyse whether OsMS188 and MS188 activate expression of OsTMS18 and AtTMS18. We generated constructs containing the coding DNA sequence (CDS) of the two MYB genes driven by the cauliflower mosaic virus (CaMV35S) promoter (p35s::OsMS188nos or p35s::MS188nos) as an effector; the firefly luciferase (LUC) reporter gene driven by the OsTMS18 or AtTMS18 promoters together with the 35S::Renilla gene served as the reporter (pOsTMS18::LUC; pAtTMS18::LUC). A construct containing only the 35S promoter and the terminator was used as a negative control (p35s::nos). As the bHLH transcription factors TDR and AMS are upstream regulators of OsMS188/AtTMS18 (Han et al., 2021), we also generated p35s::TDRnos and p35s::AMSnos constructs as other effectors. When the MYB effector and reporter were cotransformed into protoplasts, LUC luminescence was significantly enhanced compared with the background level in the negative control (Figure 5c,d). These results indicate that the MYB transcription factors play a key role in activating the expression of OsTMS18 and AtTMS18 in rice and Arabidopsis (Figure 5e,f).

Figure 5

The OsTMS18 and AtTMS18 are regulated by OsMS188 and MS188 respectively. (a) The promoter region and the MYB binding sites (black dots) of OsTMS18. The black lines show the probes for the EMSA assay. EMSA assay showed OsMS188 could bind to probes in vitro; (b) The MYB binding sites (black dots) of AtTMS18. The black lines show the probes for the EMSA assay. EMSA assay showed MS188 could bind to probes in vitro. The first lane represents free probe, and the last lane indicates the mixture of free probe and MBP tag; both are used as negative controls. The shift band is indicated by the arrowhead, which is highlighted by the positive control of mixture of biotin‐tagged probe and non‐biotin‐tagged probe and MBP‐OsMS188/ MBP‐MS188. (c, d) The transient dual‐luciferase assays were conducted in the Arabidopsis protoplasts. (c) p35s::OsMS188 and p35s::TDR were cotransformed with pOsTMS18::LUC, respectively. (d) p35s:: MS188 and p35s::AMS were cotransformed with pAtTMS18::LUC, respectively. Three replicates were performed, and the Y‐axis is shown as the ratio of Luciferase/Renilla. SD is indicated as error bar. (e, f) The tapetal genetic pathway including OsMS188 and MS188 regulate the pollen wall formation via OsTMS18 and AtTMS18 in rice and Arabidopsis, respectively. [Colour figure can be viewed at wileyonlinelibrary.com]

The application potential value of ostms18 for rice two‐line hybrid breeding

To verify whether ostms18 has a potential in breeding applications, we crossed the ostms18 line with different indica rice accessions to create male sterile lines. The F2 progeny lines with ostms18 all exhibited a male‐sterile phenotype during the high‐temperature season, suggesting that male sterility due to ostms18 would likely work well in diverse genetic backgrounds (Figure S9); multigeneration backcrossing is being implemented to generate introgression lines.

Moreover, we created an indica‐japonica hybrid by using the male sterile line ostms18 in the japonica ZH11 background as a female parent to cross with A233, an indica variety harbouring the indica‐japonica wide compatibility gene S5 (Yang et al., 2012). We analysed the agronomic traits of the F1 hybrids in a Shanghai paddy field (Figure S10). Statistical analysis showed that japonica ZH11 (ostms18) × indica A233 (OsTMS18 and wide compatibility S5) F1 plants performed similarly or better than their parents in terms of leaf length, leaf width and plant height, indicating that their vegetative growth showed heterosis (Figure S10d–f). The number of tillers in the A233 cultivar was far more than that in the ZH11 cultivar, though the seed number per panicle of A233 was slightly less than that of ZH11. The number of tillers of F1 hybrids was similar to that of A233 (Figure S10g), and the seed number per panicle was close to that of ZH11 parent (Figure S10h). Therefore, these F1 hybrids outperformed their parents in seed number of per‐plant (Figure S10i). Nonetheless, breeding application requires more agricultural experiments, e.g., field plot evaluation. These results indicate that the genetic resource of OsTMS18 is promising for two‐line hybrid breeding.

Discussion

The ostms18 line has the potential for rice breeding application

In this study, we obtained a novel TGMS line, ostms18, via genetic screening (Figure 1b,c). The sterility performance of the TGMS line in summer, the high‐temperature season, is critical for hybrid seed production. In agriculture, the occasional low temperature in summer leads to formation of partially mature pollen and selfing of TGMS plants, reducing the purity of hybrid seed production (Huang et al., 2014; Si et al., 2011). In 2018 and 2020, the sterility of ostms18 was stable in the high‐temperature season, and its fertility restoration efficiency in the low‐temperature season was similar to that of the TGMS lines carrying the tms5 locus (Figure S1). However, due to the occasional low temperature in summer, the seed setting of tms5 (ZH11) reached 13.33%–15.03% in 2021, whereas that of ostms18 was only 2.68%–5.85% (Figure 1g,h). After transfer into other indica cultivars, the sterility of lines containing the ostms18 locus was well maintained (Figure S9). Therefore, the sterility of ostms18 is much more stable than that of tms5 (ZH11) and is expected to have great value for rice breeding. Additionally, with the increasing occurrence of extreme weather, P/TGMS lines containing a single locus may have difficulty maintaining the stable sterility required for agricultural breeding. The sterile phenotype of P/TGMS cultivars with the pms3/ptgms12 locus shows abnormal development of tapetal cells during meiosis stages (Ding et al., 2012; Zhou et al., 2012) For TGMS lines with the tms5 locus, defects in the microspore mother cells occur before meiosis (Zhou et al., 2014). For ostms18, the defective microspore occurs at stages after meiosis (Figure 3). In general, introduction of the ostms18 locus into existing rice TGMS cultivars will prolong the period of thermal sensitivity. These integrated TGMS lines might reduce the possibility of surviving pollen and selfing in TGMS plants under occasional low‐temperature weather in summer for rice breeding. Therefore, the novel genetic locus, ostms18 is expected to be widely used in rice breeding.

OsTMS18 is involved in sporopollenin synthesis for pollen wall formation

The exine is composed of sporopollenin, which comprises an aliphatic unit and phenylpropanoid derivatives (Ariizumi and Toriyama, 2011; Xue et al., 2020). OsTMS18 encodes a member of the glucose methyl choline (GMC) oxidoreductase family (Wongnate and Chaiyen, 2013). The exine is almost absent in lines with the strong allele np1 (Chang et al., 2016; Liu et al., 2017), and ipe1, a maize mutant of the orthologous gene OsTMS18 (Chen et al., 2017). Therefore, OsTMS18/NP1 is essential for pollen wall formation. Members of the GMC oxidoreductase family catalyse oxidation of an alcohol moiety to the corresponding aldehyde (Kurdyukov et al., 2006; Wongnate and Chaiyen, 2013). OsTMS18 may convert fatty alcohols to fatty aldehydes for aliphatic unit synthesis. It was reported that the catalysis products of GMC oxidoreductase include H2O2 (Kurdyukov et al., 2006; Wongnate and Chaiyen, 2013), though it is not clear how the aliphatic unit and phenylpropanoid derivatives are further polymerized into sporopollenin. Recent studies have suggested that ROS are required for assembly of sporopollenin precursors (Xie et al., 2014), and it is likely that OsTMS18 also participates in the assembly of the exine layer. Sporophytic tapetal cells provide sporopollenin precursors for pollen wall formation. In both rice and Arabidopsis, a genetic pathway exists to regulate sporopollenin synthesis for pollen wall formation (Cai et al., 2015; Shi et al., 2015; Zhu et al., 2011). In this pathway, the transcription factor OsMS188/MS188 directly activates the expression of most sporopollenin‐related genes (Han et al., 2021; Wang et al., 2018; Xiong et al., 2016). Our results demonstrate that OsTMS18 and AtTMS18 are the direct targets of OsMS188 and MS188, respectively, further supporting their involvement in sporopollenin synthesis (Figure 5). This indicates that this regulatory pathway is conserved between monocotyledons and dicotyledons.

allows the creation of novel P/TGMS lines by gene editing

The GMC oxidoreductase family proteins in fungi contain the ADP‐binding motif at the N‐terminal, FAD‐binding motif at the internal and the enzyme active site at the C‐terminal regions (Zámocký et al., 2004). The mutation in np1‐1 occurs in the C‐terminal enzymatic catalysis region of OsTMS18/NP1, which leads to abnormal vacuolization of the tapetum and a lack of sporopollenin precursor. Therefore, the substrate catalysis region at the C‐terminus is essential for OsTMS18/NP1 function (Chang et al., 2016). np1‐2, np1‐3 and np1‐4 are mutations resulting from T‐DNA insertion in the middle region, a 1‐bp insertion in the C‐terminal region, and an 8‐bp deletion in the N‐terminal region, respectively. All these mutations lead to frame shifts that completely destroy the C‐terminal substrate catalysis region (Chang et al., 2016; Liu et al., 2017). These strong alleles of NP1 exhibit smaller and shrunken anthers without pollen grains, but their sterility is not affected by temperature and photoperiod (Chang et al., 2016; Liu et al., 2017). In the ostms18 mutant, a point mutation occurs in a conserved region of the N‐terminal ADP binding site (Figure 2b). Cytological analysis showed that this mutation does not affect tapetal development at high temperature; thus, the sporopollenin precursor might be partially synthesized, with formation of an exine‐like structure (Figure 3). It is likely that ostms18 is a weak allele. mOsTMS18 (G61S) may still have some catalytic activity for sporopollenin synthesis. This allows ostms18 to form an exine‐like structure, which provides a cellular basis for fertility restoration. The CRISPR system has been successfully used to edit rice genes. It is possible to create more weak alleles using this system to alter the ADP binding site or other regions in OsTMS18/NP1. Further screening of these weak alleles would provide an opportunity to obtain novel TGMS lines in rice.

Cellular mechanism for fertility restoration of P/TGMS lines in plants

During anther development, the structure and composition of cell walls from sporogenous cells to pollen undergo tremendous changes. After meiosis, microspores are enclosed by tetrad walls. With gradual degradation of this wall, the released microspore is protected by the outer pollen wall exine. Later, the intine layer is formed between the exine and plasma membrane to protect the mature pollen. In Arabidopsis thaliana, multiple TGMS genetic loci have been reported to participate during this process, such as CalS5, RVMS, RPG1, ACOS5, CYP703A2, ABCG26 and MGT5 (Xu et al., 2015; Xu et al., 2020;Zhang et al., 2020; Zhu et al., 2020). These TGMS genes in Arabidopsis are directly or indirectly involved in cell wall transition from the tetrad wall to the pollen wall. In the present work, both OsTMS18 and AtTMS18 were found to be related to exine formation (Figures 3 and S7). Among the reported TGMS genetic loci, the tms5 mutant exhibits abnormal MMC development (Zhou et al., 2014), which might affect tetrad wall formation. Mutation of noncoding RNAs in pms3/ptms12 and LRR receptor kinase in tms10 results in irregular tapetal development, which affects pollen wall formation (Ding et al., 2012; He et al., 2001; Yu et al., 2017). Hence, these P/TGMS genes in rice are also related to cell wall formation during microspore development into pollen (Figure 6a). However, the reported P/TGMS genetic loci in rice are quite different from those in Arabidopsis. In general, rice grows in paddy fields, with its reproductive stage under strong sunshine and high temperatures in summer. The totally inadequate exine with the strong alleles np1 fail to protect developing microspores (Chang et al., 2016; Liu et al., 2017), whereas the partially synthesized sporopollenin wall with the weak allele ostms18 might protect the microspore, allowing it to develop into functional pollen under low‐temperature conditions (Figure 3). As a shade plant, Arabidopsis thaliana can be cultured in the laboratory, requiring lower levels of wall protection than rice. This allows for sufficient protection in attms18 for microspore development under the typical growth temperature (24°C) (Figure 4c,d). In contrast, attms18 does not have enough protection under high temperature (28°C) which results in serious sterility (Figure 4e,f).

Figure 6

A proposed model of a cellular mechanism for plant P/TGMS fertility restoration. (a) When the meiosis completed, the haploid microspores are enclosed inside tetrad wall, which protects microspores and facilitates exine precursors deposit onto microspores. Then tetrad wall is degraded to release microspores, and exine layer takes over the protective role for microspores. The intine formation is formed at late stages to carry on the protection. Related genes were listed on the bottom of each event. (b) The proposed model for the fertility restoration of a P/TGMS line. The essence is whether the protection provided by the P/TGMS line could meet the requirement of its microspore development under different environment conditions. The protection provided by single P/TGMS line is constant. However, different growth environment and plant species are the variable parameters, requiring different strength of protection and leading to sterility or fertility. [Colour figure can be viewed at wileyonlinelibrary.com]

Slow development is a general mechanism for restoring the fertility of P/TGMS lines under low temperatures or short‐day photoperiods in Arabidopsis (Zhang et al., 2020; Zhu et al., 2020). The identification of restorers of the TGMS line rvms demonstrates that slow development reduces the requirement for cell wall protection. In this case, the defective walls of P/TGMS lines are sufficient to support microspore development into functional pollen (Shi et al., 2021; Wang et al., 2021). The fertility of a P/TGMS line depends on whether the protection provided meets the requirement of microspore development under different environmental conditions. We propose a cellular mechanism for fertility restoration of TGMS lines that are integrated with genetic mutations and different plant species (Figure 6b). (1) For a specific P/TGMS gene, different genetic mutations may provide varying levels of protection for microspore development. For a line with a specific mutation, the wall protection is determined. (2) Different plant species, such as rice and Arabidopsis, require different levels of protection for microspore development. (3) Under slow‐growth conditions, including low temperatures or short‐day photoperiods, the requirement for cell wall protection is reduced. The sterility or fertility of P/TGMS lines is dependent on multiple factors, including different genetic mutations in a specific gene, different plant species and their growth environments (Figure 6b). This general understanding may facilitate the creation of novel P/TGMS lines and guide crop breeding in agriculture.

Materials and methods

Plant materials

The rice lines used in this study were ostms18 (an EMS mutagenesis‐induced japonica TGMS line), wildtype ZH11 (japonica), Kasalath (indica) and A233 (indica). These lines were grown in a botanical garden and paddy field in Shanghai. The F2 populations were generated by crosses between ostms18 and Kasalath for first‐pass mapping and ostms18 and wildtype ZH11 for BSA‐seq analysis. The Arabidopsis (Arabidopsis thaliana) Colombia ecotype was used as wildtype. Seeds were placed in the dark at 4°C for 2 d before germination and then planted in a 24°C incubator with 16 h light and 8 h dark. The generation of attms18–1 was mediated by CRISPR‐Cas9 technology. The T‐DNA line of AtTMS18 (attms18–2, SALK_031400) was purchased from the Salk mutant library (http://signal.salk.edu/).

Phenotypic analysis

The plants were imaged with a Nikon D7000 digital camera (Nikon, Tokyo, Japan) and an Olympus SZX10 dissecting microscope (Olympus, Tokyo, Japan). Anthers were captured with an Olympus BX51 fluorescence microscope (Olympus, Tokyo, Japan). Pollen viability was assessed using Alexander's solution as previously described (Lou et al., 2014). The cytological characteristics of all materials were observed by semithin sectioning, TEM (JEOL, Tokyo, Japan) and SEM (Hitachi, Tokyo, Japan). Embedding and observation procedures were performed as described in a previous study (Lou et al., 2014).

Analysis of TGMS traits

Each batch of ostms18, AnS‐1 and tms5 (ZH11) seeds was germinated once a week since May 18th (in 2018 and 2020), and seedlings were transplanted to the paddy field when the plant height was approximately 20 cm. To analyse the relationship between the booting temperature and seed‐setting rate, the average temperature was recorded when the plants entered the booting stage, and spikes were collected to assess the seed‐setting rate of different batches.

Complementary analysis and protein expression

For complementary analysis, the full‐length genome of OsTMS18 was cloned into the binary vector pCAMBIA1300 driven by the native promoter. For protein expression, OsTMS18‐GFP and AtTMS18‐GFP binary vectors were constructed by fusing GFP with the CDS of OsTMS18 and AtTMS18 driven by their native promoters. These constructed plasmids were transferred into A. tumefaciens and then used to infect Arabidopsis flower buds. T1 seeds were selected using 20 mg/L hygromycin. Rice genetic transformation was performed by Boyuan Biological Company (Boyuan, Wuhan, China). An Olympus FV3000 laser confocal microscope was used to observe GFP green fluorescence (Olympus, Tokyo, Japan).

Expression analysis

The RNA used for RT‐PCR analyses was derived from the inflorescences of wild type, udt1, ostdf1, tdr, osms188, bhlh142 and eat1 plants using a TRIzol kit (Invitrogen, Carlsbad, CA, USA). Reverse transcription was performed using TransScript Fly First‐Strand cDNA Synthesis SuperMix (TransGen Biotech, Beijing, China). The semiquantitative RT‐PCR procedure was performed as described previously (Xiong et al., 2016). qRT‐PCR analyses of each sample were performed with SYBR Green Real‐time PCR Master Mix (Toyobo, Osaka, Japan) and an ABI 7300 system (Life Technologies, Carlsbad, CA, USA). OsACTIN was used as a normalizer (for normalization corresponding to the total RNA level) for RT‐PCR assays.

Electrophoretic mobility shift assays (EMSAs)

The MBP‐OsMS188 was constructed, induced and purified as described previously (Xiong et al., 2016). Biotin‐labelled and unlabeled DNA probes were designed based on the MYB transcription factor‐binding sites in the cis‐elements of the OsTMS18 promoter (Table S1). A LightShift Chemiluminescent EMSA Kit (Thermo Scientific, Waltham, Massachusetts, USA) was used to perform the EMSAs. Finally, images were taken with a Tanon‐5500 Chemiluminescent Imaging System (Tanon, Shanghai, China).

Dual‐luciferase transient expression assays of Arabidopsis protoplasts

The promoter of OsTMS18 was cloned into the pGreenII 0800‐LUC vector. Cellulase (0.015 g/mL) and pectinase (0.0035 g/mL) were used to treat Arabidopsis (Col‐0) leaves grown for 21–28 days to obtain Arabidopsis protoplasts. The plasmids p35S::TDR‐nos, p35S::OsMS188‐nos and pOsTMS18::LUC were cotransfected into the protoplasts in polyethylene glycol solution (0.4 g/mL PEG4000), and the cytoplasm was cultivated with 12–16 h of light treatment. The dual‐luciferase reporter assay was performed as described previously (Wang et al., 2018).

BSA sequencing

F2 generation plants derived from ostms18 and wild‐type ZH11 were separated into two pools: fertile lines as one pool and sterile lines as the other. Genomic DNA was extracted from the rosette leaves of 70 lines from each pool and their parents. Total DNA was randomly digested into small fragments, which were subsequently purified by electrophoresis for library construction. The library was prepared and sequenced by Institute of Plant Physiology & Ecology, CAS. The filtering process and statistical tests for sequencing data have been described previously (Wang et al., 2019).

Phylogenetic analysis

The full‐length protein sequence of OsTMS18 was used to search for homologous sequences in Basic Local Alignment Search Tool (BLAST) of the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/). Multiple sequence alignment was performed using the ClustalW tool online (http://www.ch.embnet.org/software/clustalW.html) and displayed using BOXSHADE (http://www.ch.embnet.org/software/BOX_form.html). Phylogenetic trees were constructed and tested by MEGA3.1 based on the neighbour‐joining method.

Conflict of interest statement

The authors declare no competing interests.

Author contributions

The project leader: Z.‐N.Y. Conceived the experiments: Z.‐N.Y. and J.Z. Phenotype and cytological analysis: Y.‐F.Z. and Y.‐L.L. Genetic complementation: Y.‐L.L. Arabidopsis homologue analysis: Y.‐F.Z. and X.Z. Vector construction: J.‐J.W. and L.Z. Regulation pathway analysis: Y.‐F.Z., Y.H. and D.‐D. L.BAS‐seq analysis: X.‐H.H. Writing—original draft: J.Z. and Y.‐F.Z. Writing—review & editing: Z.‐N.Y., J.Z. and N.W.

Figure S1 The TGMS traits of ostms18.

Click here for additional data file.

Figure S2 The first‐pass mapping in chromosome 10 of OsTMS18.

Click here for additional data file.

Figure S3 The dehiscent anthers of WT and ostms18.

Click here for additional data file.

Figure S4 Cytological analysis of anther development in the WT and ostms18 at stages 8–9.

Click here for additional data file.

Figure S5 Expression patterns of homologue proteins of OsTMS18.

Click here for additional data file.

Figure S6 Identification of AtTMS18.

Click here for additional data file.

Figure S7 SEM analysis of pollen grains in attms18 mutants.

Click here for additional data file.

Figure S8 Expression of OsTMS18 in transcription factors' mutants.

Click here for additional data file.

Figure S9 The phenotype ostms18 in different Indica rice germplasms.

Click here for additional data file.

Figure S10 The heterosis of F1 hybrids using the ostms18 crossed with Indica rice.

Click here for additional data file.

Table S1 List of Primers used in the Study and Their Sequences.

Click here for additional data file.