The mitochondrial aldehyde dehydrogenase OsALDH2b negatively regulates tapetum degeneration in rice

Rice mitochondrial aldehyde dehydrogenase OsALDH2b contributes to tapetal degradation, which is essential for male fertility.

Rice has 22 aldehyde dehydrogenase members grouped into 11 families, including OsALDH7, OsALDH2a, and OsALDH2b (Gao and Han, 2009). In this study, we identified a rice male-sterility mutation caused by a 7-bp deletion in OsALDH2b. Furthermore, we revealed that OsALDH2b encodes a mitochondrion-targeted aldehyde dehydrogenase enzyme and is highly expressed in anthers during microsporogenesis. Our results demonstrate that OsALDH2b removes excess aldehydes generated during anther development to negatively regulate tapetum degeneration.

Plant materials
All rice plants were grown under natural conditions in South China Agricultural University at Guangzhou's paddy field. The male-sterile mutant (later named osaldh2b) was obtained from a 60 Co-γ-ray-treated rice Nipponbare (O. sativa, ssp. japonica) mutant library. The F 2 mapping population was generated from a cross between the mutant and an indica variety, Huanghuazhan (HHZ, O. sativa, ssp, indica). In the F 2 population, male-sterile plants were selected primarily for genetic mapping. For screening recombinant individuals, F 2 and F 3 segregants were planted in 96-well plates and used for high-throughput DNA preparation as described previously .

Mutant phenotype characterization
Plants were photographed with a Nikon digital camera. Flowers were photographed with a stereomicroscope (SZx10/DP72, Olympus, Japan). Pollen grains were stained with 1% I 2 -KI solution and photographed with a fluorescence microscope (Axio Observer Z1, Zeiss, Germany). Preparation of rice anther sections for light microscopy and electron microscopy was performed as previously described .

Map-based cloning of OsALDH2b
A set of 145 male-sterile plants segregated from 652 F 2 individuals was used for primary mapping. Recombinants were then screened from F 2 and F 3 families for fine mapping with newly developed insertion/deletion (InDel) molecular markers (see Supplementary Table S1 at JXB online). Rice genomic DNA samples were prepared from fresh leaf tissues using 1% sodium dodecyl sulfate.

Vector construction for transgenic plants
For the functional complementation test, a 12.2 kb wild-type genomic fragment of OsALDH2b was amplified by three steps. The first fragment was amplified using OsALDH2b-F1 and OsALDH2b-R1 primers (Supplementary Table S2) and cloned into the MluI and SalI sites of the binary vector pCAMBIA1300.2. The second fragment was amplified with OsALDH2b-T5F2 and OsALDH2b-T5R2 primers and cloned, into the positive clones produced in the first step, at the SalI site using an isothermal in vitro recombination (IR) system (Jiang et al., 2013). The third fragment, containing the 5′-upstream region, was amplified with pALDH2b-T5F and pALDH2b-T5R primers, and inserted into the vector constructed in step 2 at the BamHI site by the IR method. The CRISPR/Cas9 genome-targeting construct for OsALDH2b (target site: TGGGACACAAGGATTGTTGCCGG; protospacer adjacent motif italicized) was designed with the web-based CRISPR-GE toolkit (http://skl.scau.edu.cn/) (Xie et al., 2017) and prepared using the CRISPR/Cas9 vector system . All constructs were introduced into rice with Agrobacterium-mediated transformation. Positive transformants were screened with HPT primers by PCR. The target site sequences of gene knockout mutants were sequenced and decoded with CRISPR-GE/DSDecodeM (Liu et al., 2015;Xie et al., 2017).

RNA extraction and qRT-PCR
Total RNA was extracted from rice tissues using TRIZOL reagent (Thermo Fisher Scientific, USA), and isolated RNA was treated with DNase I. The treated RNA was then used for first-strand cDNA synthesis with oligo (dT) using the first-strand cDNA synthesis kit (Promega, USA). Two microliters of the reverse transcription product was used as the template for PCR reactions. The quantitative reverse transcription polymerase chain reaction (qRT-PCR) of OsALDH2b and other genes related to anther development used the primers listed in Supplementary Table S2.

Subcellular localization
The coding region of OsALDH2b was amplified from wild-type cDNA with the primers OsALDH2b-cF and OsALDH2b-cR (Supplementary  Table S2). After digestion with HindIII and BamHI, the fragments were fused in-frame with the enhanced green fluorescent protein (eGFP) coding sequence (Heim et al., 1995;Cormack et al., 1996), subcloned into a pUC-18-based vector and driven by the CaMV35S promoter to produce the transient expression vector ALDH-eGFP. A mutant orange fluorescent protein (mOrange) fused with a mitochondrial transit signal peptide derived from RF1b (Wang et al., 2006) was prepared as a positive control (RF1b-mOrange). These constructs were bombarded into onion epidermal cells by a helium-driven accelerator (PDS/1000; Bio-Rad, USA). Cells that exhibited eGFP and mOrange fluorescence were imaged with a laser scanning confocal microscope (LSM7 DUO, Zeiss, Germany).

Aldehyde dehydrogenase enzymatic assays
Full-length OsALDH2b cDNA (excluding the mitochondriatargeted sequence) was isolated with the primers OsALDH2b-cFD and OsALDH2b-cR (Supplementary Table S2) and cloned into the pET32a(+) vector fused with a His-tag. The resultant plasmid was transformed into E. coli strain BL21 (DE3). Once the OD 600 reached approximately 0.6, transformed cells were incubated at 18 °C for 16 h with 1 mM isopropylthio-β-galactoside. The supernatant containing extracted proteins was purified with a Ni-nitrilotriacetic acid spin column. For the enzymatic assay, aldehyde was the substrate. ALDH enzymatic activity for reduction of NAD + to NADH was evaluated by the increase of absorbance at 340 nm (Shin et al., 2009).

Measurement of malonaldehyde content in the anthers
Determination of MDA levels was by the thiobarbituric acid (TBA) method (Loreto and Velikova, 2001). Anther samples at different developmental stages (each 100 mg) were homogenized in 2 ml of 0.1% trichloroacetic acid solution, and the extract was centrifuged at 12 000 g for 15 min; 0.5 ml of the supernatant was diluted to 1 ml with 0.5% TBA in 20% trichloroacetic acid. The mixture was heated at 95℃ for 30 min and then cooled on ice. Supernatant absorbance was measured at 530 nm with a Synergy Mx Multi-Mode Reader (BioTek, USA), subtracting non-specific absorbance at 600 nm.

TUNEL assay
Anther developmental stages were confirmed by observing anther crosssections with light microscopy. Preparation of anther sections and a terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) assay used a Dead End Fluorometric TUNEL Kit (Promega, USA); these were performed as previously described Luo et al., 2013). Fluorescein's green fluorescence (TUNEL signal) and propidium iodide's red fluorescence were imaged with 488 nm (excitation) and 520 nm (detection), and 488 nm (excitation) and 610 nm (detection), respectively, under a LSM 7 DUO laser scanning confocal microscope (Zeiss, Germany).

Identification and phenotype of the osaldh2b mutant
We obtained a male-sterile mutant by screening a rice mutant library (japonica cultivar Nipponbare) created with 60 Co-γ-ray radiation. The mutant exhibited normal vegetative and panicle development, but failed to generate viable pollen, and never set seed ( Fig. 1A-D). Analysis revealed that the sterility was caused by a loss-of-function mutation located in the OsALDH2b gene (see below); we therefore named this mutant osaldh2b. We crossed mutant plants with pollen grains from the indica variety (HHZ) to generate F 1 hybrids. We examined 652 F 2 individual plants that resulted from the cross, and observed that 507 plants were male fertile and 145 male sterile (χ 2 =2.65 for 3:1, P>0.05) (Supplementary Table S3), indicating that the malesterility phenotype was controlled by a recessive locus.

Map-based cloning and functional validation of OsALDH2b
To isolate the mutated gene conferring male sterility, we used 6652 segregants from the F 2 and F 3 families and a set of polymorphic markers covering the entire genome, and mapped the mutant locus to a 114-kb region on chromosome 6 ( Fig. 2A). DNA sequencing analysis in this region of the mutant revealed a 7-bp deletion in the third exon of OsALDH2b (LOC_Os06g15990, annotated by RGAP7; or Os06g0270900, annotated by RAP-DB), causing a frame shift to introduce a premature stop codon at the 125th amino acid (Fig. 2B).
To confirm that the male-sterile phenotype resulted from the mutation in OsALDH2b, we prepared a binary construct (OsALDH2b-C) that carried a 12.2-kb genomic DNA fragment of wild-type OsALDH2b to thoroughly test functional complementation of this fragment comprising OsALDH2b's 4.2-kb upstream 5′-UTR/promoter sequence, the entire 5.4kb coding region (including introns), and a 2.6-kb downstream region. We transformed this construct into calli induced from heterozygous OsALDH2b/osaldh2b plants. Of 26 OsALDH2b-C transgenic plants (T 0 ), six plants were homozygous for the osaldh2b allele, and these exhibited normal male fertility (Fig. 2C). Next, we used CRISPR/Cas9-based genome editing to knock out OsALDH2b in wild-type plants. As expected, the OsALDH2b-knockout plants (OsALDH2b-KO) exhibited a male-sterile phenotype, similar to the osaldh2b mutant ( Fig. 2C; Supplementary Fig. S1). Therefore, we concluded that OsALDH2b is required for male development in rice.

OsALDH2b encodes a mitochondrial aldehyde dehydrogenase and is highly expressed in anthers
Sequence analysis showed that OsALDH2b encodes a predicted 549 amino acid mitochondrial aldehyde dehydrogenase ( Supplementary Fig. S3). To verify OsALDH2b's subcellular localization, we co-transformed OsALDH2b-eGFP and an RF1b-mOrange control into onion epidermal cells. Images demonstrated that the OsALDH2b-eGFP signal co-localized with RF1b-mOrange in mitochondria, indicating that OsALDH2b is a mitochondrion-localized protein (Fig. 4A).
To examine whether OsALDH2b has enzymatic activity for aldehydes, we purified recombinant OsALDH2b-His to use in enzyme activity assays. Here, reduction of NAD + to NADH was measured as the increase in absorbance values at A 340 . When aldehyde was used as the substrate, recombinant  OsALDH2b-His exhibited significant enzymatic activity (Fig. 4B), demonstrating that the protein has aldehyde dehydrogenase activity.
To investigate OsALDH2b function, we analysed its expression pattern during rice development with qRT-PCR. OsALDH2b was expressed in both vegetative and reproductive organs. During anther development, OsALDH2b was highly expressed from the meiosis I stage (S7) until the middle microspore stage (S9b), and peaked at the meiosis II/tetrad (S8) and early microspore (S9a) stages (Fig. 4C); rice anther stages (S1-S12) were assigned as previously described . OsALDH2b's expression profile was consistent with transcriptome data (RiceXPro, http://ricexpro.dna.affrc.go.jp/) ( Supplementary Fig. S4). These data suggest that OsALDH2b may function in anther and pollen development from meiosis to the microspore stages.

The osaldh2b mutant anthers accumulate excess malonaldehyde
Cellular redox state is a key factor for male gametogenesis (Zhang and Yang, 2014). MDA is the predominant product of oxidative stress and one of the most highly reactive of the endogenous aldehydes, which are triggered by hypoxic status during early anther development and potentially produce toxic byproducts (Voulgaridou et al., 2011). To determine whether OsALDH2b acts in anther development by reducing the aldehyde accumulation, we used a TBA assay to measure MDA content in anthers of both wild-type and osaldh2b plants.
In wild-type anthers, MDA content gradually increased from the microspore mother cell stage (S6) to meiosis/tetrad stages (S7/S8), and then decreased until the late microspore stage (S10) (Fig. 5). In osaldh2b anthers, however, the MDA level was much higher from the S8 to S10 stages. This result indicated that OsALDH2b acts as a detoxifying enzyme that eliminates aldehydes generated during anther and microspore development.  The osaldh2b mutant exhibits premature tapetal programmed cell death MDA is a highly reactive aldehyde that reacts strongly with DNA and proteins (Voulgaridou et al., 2011). To examine whether MDA accumulation affects DNA fragmentation in osaldh2b anthers, we performed a TUNEL assay on anthers across developmental stages. Wild-type anthers showed strong TUNEL-positive signals in tapetal cells at the tetrad stage (S8b) (Fig. 6A-E, top). However, in osaldh2b tapetal cells, we detected TUNEL-positive signals in earlier stages, particularly in the metaphase I stage (S8a) (Fig. 6A-E, bottom). These results demonstrate that PCD-induced tapetal DNA fragmentation occurred at an earlier time point in the osaldh2b mutant, suggesting that excess MDA accumulation in the mutant's developing anthers may accelerate PCD in tapetum cells.

The osaldh2b mutant exhibits abnormal tapetal degeneration and microspore development
To further investigate the role of OsALDH2b during male reproductive development, we analysed semi-thin sections of wild-type and mutant anthers. We observed no obvious differences between cells in wild-type and osaldh2b at early developmental stages (microspore mother cell stage to the early microspore stage) (Fig. 7A-D). In both wild-type and osaldh2b anthers, microsporocytes and somatic layers (including the epidermis, endothecium, middle layer, and tapetum) exhibited characteristic structures (Fig. 7A). Microsporocytes in both wild-type and osaldh2b had progressed through normal meiosis (S7), during which the tapetum had become vacuolated (Fig. 7B); subsequently, tetrads of haploid microspores had formed (S8) (Fig. 7C). At the early microspore stage (S9a) in both genotypes, free microspores had been released from tetrads, the middle layer appeared thin, and the tapetum looked condensed, less vacuolated, and deeply stained (Fig. 7D).
We detected morphological differences between osaldh2b and wild-type anthers starting from the middle microspore stage (S9b): the wild-type tapetum was evident and microspores were round and vacuolated (Fig. 7E, top). The osaldh2b tapetum, however, appeared thinner, and microspores appeared irregularly shaped (Fig. 7E, bottom). At the late microspore stage (S10), wild-type tapetum had become hill-shaped, appeared highly condensed and deeply stained, and formed microspores containing a single, large central vacuole (Fig. 7F, top). By contrast, osaldh2b tapetum was less condensed and weakly stained; microspores appeared collapsed and exhibited uneven cytoplasm associated with abnormal vacuolization (Fig. 7F,  bottom). At the bicellular pollen stage (S11), wild-type anthers exhibited typical falcate-shaped pollen grains and completely Fig. 5. MDA content dynamics during anther development. MDA content of osaldh2b anthers was significantly higher than that in WT anthers at microspore stages (S9 and S10). These data are derived from three replicates; **P<0.01 by t-test. Fig. 6. Tapetal nuclear DNA fragmentation in WT and osaldh2b anthers. The anthers in WT (top) and osaldh2b (bottom) from the microspore mother cell stage through the middle microspore stage were compared for nuclear DNA fragmentation (indicating PCD) using the TUNEL assay (A-E). Nuclei were stained with propidium iodide (red fluorescence); yellow signals indicate TUNEL-positive nucleus staining. MMC, microspore mother cell; T, tapetum; Msp, microspore. Scale bars: 50 µm. degenerated tapetal cells (Fig. 7G, top). At this stage, osaldh2b anther wall layers, including the epidermis and endothecium, appeared disordered, enlarged, and broken; mutant plants had produced severely aberrant microspores (Fig. 7G, bottom). At the mature pollen stage (S12), in contrast to wild-type, osaldh2b pollen grains were irregularly shaped and had accumulated no or less storage materials, and the anthers had shriveled (Fig. 7H).
We used transmission electron microscopy to study the developmental abnormalities at the microspore stages in more detail. At the early microspore stage (S9a), wild-type tapetal cytoplasm was highly condensed, nuclei were intact, and cells exhibited a prominent nucleolus (Fig. 8A, top). Strikingly, we observed no nucleolus in osaldh2b tapetum nuclei at this stage (Fig. 8A, bottom). At the middle microspore stage (S9b), the wild-type tapetum had collapsed and nuclei were lobed. Enlarged U-shaped orbicules were evident on the inner tapetal surface. The exine in wild-type microspores was well established with distinct nexine, tectum, and bacula layers (Fig. 8B,  top). By contrast, at this stage in osaldh2b tapetal cells, nuclei appeared completely degenerated, orbicules were smaller, and electron-dense sporopollenin reduced. Moreover, the exine of osaldh2b microspores was much thinner compared with those in WT. The osaldh2b microspores contained few organelles in cytoplasm observed in electron-transparent channels (Fig. 8B, bottom). At the late microspore stage (S10) in wildtype, we observed further degenerated tapetum and vacuolated Fig. 7. Transverse section analysis reveals anther development in WT and osaldh2b. (A-D) No obvious differences were observed between WT (top) and osaldh2b (bottom) anthers from S6 to S9a. (E) Compared with WT anthers, osaldh2b anthers displayed thinner tapetum and irregular microspores in the locule at S9b. (F) The osaldh2b tapetum was less condensed and weakly stained, and microspores were collapsed with uneven cytoplasm at S10. (G) The osaldh2b anther wall layers, including epidermis and endothecium, appeared disordered, enlarged, and broken, and exhibited severely abnormal microspores at S11. (H) The osaldh2b anthers at S12 exhibited collapsed pollen grains with no or less cellular content accumulation. E, epidermis; En, endothecium; ML, middle layer; Mp, mature pollen; Ms, microsporocyte; Msp, microspores; T, tapetum; Tds, tetrads. Scale bars: 20 μm. microspores with abundant cytoplasm (Fig. 8C, top). At this stage, osaldh2b tapetum exhibited cavities, with low-electrondensity orbicules on its surface, indicating that its tapetum had completely and prematurely degraded. In addition, microspore exine was much thinner (Fig. 8C, bottom). We further investigated the expression of eight genes related to male reproductive development. The qRT-PCR results ( Supplementary  Fig. S5) showed that the mutation of OsALDH2b disrupted the expression of genes involved in tapetum degeneration (TDR,UDT1,OsGAMYB,RTS) and pollen wall formation (WDA1, CYP704B2, CYP703A3, OsC6).
Together, these results suggested that defective mutation in OsALDH2b leads to excessive aldehyde accumulation, which causes early tapetal PCD, premature cellular degeneration, and aborted microspore development, resulting in male sterility (Fig. 9).

Discussion
The tapetum is arguably the most important layer of anther tissue during male meiosis and microsporogenesis, providing enzymes, signals, and nutrients for pollen development via PCD-based cellular degeneration Guo and Liu, 2012). Many components and factors participate in the process of tapetum development, such as transcription factors, receptor-like kinases, and transporters. The bHLH transcription factors TDR, EAT1/DTD, and TIP2 function as crucial positive regulators to promote tapetal PCD Ji et al., 2013;Niu et al., 2013;Fu et al., 2014). Mutation of their genes leads to vacuolated and prematurely degraded tapetum. In this study, we identified a rice male sterility mutant, osaldh2b; its wild-type gene encodes a conserved mitochondrial aldehyde dehydrogenase, OsALDH2b (Figs 1-4). Cytological analysis showed that osaldh2b exhibited more rapid, prominent degradation of tapetal cell nuclei and formation of abnormal tapetal secretory structures at microspore stages (Figs 7-8). Consistent with the nucleus degradation, tapetum DNA fragmentation (indicating PCD) occurred earlier in the osaldh2b mutant, at the prophase I stage (Fig. 6). Therefore we infer that OsALDH2b plays an important role in anther development and pollen formation by negatively regulating tapetal PCD. Additionally, expression analysis of some marker genes related to anther development indicates that the defective OsALDH2b caused disorder of the regulatory networks for anther development ( Supplementary Fig. S5). The expression of TDR is increased from meiosis to microspore stages in osaldh2b. As the function of TDR is to promote the initiation of tapetal PCD , the up-regulated change of TDR expression is consistent with the earlier occurrence of tapetal PCD in osaldh2b anthers. Furthermore, the expression of GAMYB, Fig. 8. Transmission electron micrographs of WT and osaldh2b anthers. Tapetum and microspores in WT (top) and osaldh2b (bottom) anthers from S9a to S10 are shown. (A) In WT tapetum at S9a, nuclei appeared intact with a visible nucleolus (arrowed); the nucleolus had disappeared in the osaldh2b tapetum at this stage. (B) At S9b, nuclei in osaldh2b tapetal cells were completely degenerated, and mutant orbicules were smaller with reduced electron-dense sporopollenin, and the exine (including tectum, bacula, and nexine) of osaldh2b microspores was thinner than in WT. (C) At S10, the osaldh2b tapetum had become a cavity, with low-electron-density orbicules on its surface, indicating complete and premature tapetum degradation; the exine of osaldh2b microspores appeared thinner, with barren cytoplasm compared to WT. Ba, bacula; Ex, exine; Msp, microspores; Ne, nexine; Nu, nucleus; Or, orbicule; T, tapetum; Te, tectum. Scale bars: 1 µm.
which is involved in the down-regulation of TDR expression in anthers (Aya et al., 2009;Liu et al., 2010), is decreased in the mutant. According to the expression analysis, it seems that TDR may act downstream of OsALDH2b in regulating tapetal PCD, but this needs further investigation.
Dynamic redox status is an emerging factor affecting tapetum specification and timing degradation. Two Cys-rich metallothioneins, OsMT2b and OsMT-I-4b, have been identified as ROS scavengers. DTC1 interacts with OsMT2b and inhibits the ROS scavenging activity of OsMT2b to ensure timely production of ROS for proper initiation of tapetal PCD during early stage anther development (Yi et al., 2016). On the contrary, OsMADS3 promotes the expression of MT-I-4b to eliminate the excess ROS during later stage anther development (Hu et al., 2011). Previous studies have indicated that lipid peroxidation increases upon hypoxia in plants (Blokhina and Fagerstedt, 2010;Gupta et al., 2017), resulting in production of reactive aldehydes including MDA; these are highly reactive with cellular compounds and nucleic acids (Voulgaridou et al., 2011;Heymann et al., 2018). ALDHs are major enzymes for selective elimination of aldehydes in animals and plants.
In plants, the ALDH family includes mitochondrial ALDH (mtALDH) and cytosolic ALDH (ctALDH) subgroups based on their cellular location (Kirch et al., 2004;Gao and Han, 2009;Zhou et al., 2012). The rice genome harbors two mtALDHs, OsALDH2a and OsALDH2b (Gao and Han, 2009). Here we report that OsALDH2b functions to regulate the proper levels of aldehydes during redox stress in the developing tapetum. When OsALDH2b is dysfunctional, excess aldehydes accumulate in the tapetal cells (Fig. 5). Although the significant change of MDA accumulation slightly lags behind the observed early occurrence of tapetal PCD signal in osaldh2b (Figs 5, 6), we propose that the MDA accumulation might serve as a signal to initiate the premature tapetal PCD in this mutant.
Although OsALDH2b is constitutively expressed in vegetative organs, especially in leaf, the osaldh2b mutant does not show vegetative defects. We reason that there might be functional divergence of mtALDH orthologs as described in maize (Liu and Schnable, 2002). Two mtALDHs, RF2A and RF2B, have differential accumulation and distinct enzymatic activities with their substrates; RF2A, but not RF2B, accumulates to high levels in tapetal cells and is involved in male fertility (Cui et al., 1996;Liu et al., 2001). Based on previous phylogenetic analysis of plant ALDHs, OsALDH2b is more similar to maize RF2A and OsALDH2a is more similar to RF2B (Tsuji et al., 2003). A possible role of OsALDH2a may be to eliminate acetaldehyde in vegetative tissues, so as to increase submergence tolerance (Nakazono et al., 2000). Altogether, we conclude that mtALDHs have undergone functional specialization during evolution to accommodate endogenous or exogenous stresses in different developmental organs.

Supplementary data
Supplementary data are available at JXB online. Fig. S1. CRISPR/Cas9 OsALDH2b knockout. Fig. S2. Sequence alignment of ALDH plant orthologs. Fig. S3. OsALDH2b amino acid sequence. Fig. S4. OsALDH2b expression profile based on RiceXPro. Fig. S5. Expression analysis of eight genes related to anther development. Table S1. Molecular markers used for fine mapping. Table S2. Primers used for vector construction and expression analysis. Table S3. Genetic analysis of osaldh2b. Fig. 9. Model of OsALDH2b's role in pollen development. Tapetal cells undergo high hypoxic stress at early developmental stages, triggering accumulation of reactive aldehydes (MDA). In wild-type rice, OsALDH2b converts aldehydes into acetates, which are incorporated into the TCA cycle, resulting in MDA homeostasis. Defective OsALDH2b causes excess aldehyde accumulation, which leads to premature tapetal PCD and cellular degradation, and pollen abortion. ACS, acetyl coenzyme A synthetase. TCA, tricarboxylic acid.