A detrimental mitochondrial-nuclear interaction causes ... - Nature

letters

A detrimental mitochondrial-nuclear interaction causes cytoplasmic male sterility in rice

npg

© 2013 Nature America, Inc. All rights reserved.

Dangping Luo1,2, Hong Xu1,4, Zhenlan Liu1, Jingxin Guo1, Heying Li1, Letian Chen1, Ce Fang3, Qunyu Zhang1, Mei Bai1, Nan Yao3, Hong Wu1, Hao Wu1, Chonghui Ji1, Huiqi Zheng1, Yuanling Chen1, Shan Ye1, Xiaoyu Li1, Xiucai Zhao1, Riqing Li1 & Yao-Guang Liu1 Plant cytoplasmic male sterility (CMS) results from incom patibilities between the organellar and nuclear genomes and prevents self pollination, enabling hybrid crop breeding to increase yields1–6. The Wild Abortive CMS (CMS-WA) has been exploited in the majority of ‘three-line’ hybrid rice production since the 1970s, but the molecular basis of this trait remains unknown. Here we report that a new mitochondrial gene, WA352, which originated recently in wild rice, confers CMS-WA because the protein it encodes interacts with the nuclear-encoded mitochondrial protein COX11. In CMS-WA lines, WA352 accumulates preferentially in the anther tapetum, thereby inhibiting COX11 function in peroxide metabolism and triggering premature tapetal programmed cell death and consequent pollen abortion. WA352-induced sterility can be suppressed by two restorer-of-fertility (Rf) genes, suggesting the existence of different mechanisms to counteract deleterious cytoplasmic factors. Thus, CMS-related cytoplasmic-nuclear incompatibility is driven by a detrimental interaction between a newly evolved mitochondrial gene and a conserved, essential nuclear gene. In hybrid crop breeding, crossing different inbred lines produces F1 hybrids that usually have higher yields than the parents, a phenomenon known as hybrid vigor or heterosis. However, production of sufficient quantities of hybrid seeds poses a logistical problem, as many crops, such as rice, reproduce by self pollination. Using rice CMS lines as the female parents to avoid self pollination has been crucial for the commercial production of hybrid rice seeds7,8. CMS-WA was discovered in the 1970s in a wild rice (Oryza rufipogon), and this cytoplasm was backcrossed into indica rice (Oryza sativa ssp. indica) to produce CMS-WA lines7. Hybrid rice bred by the three-line (CMS, maintainer and restorer) technology, ~99% of which use CMS-WA lines and other CMS lines that also carry the same CMS-WA gene, has strong heterosis that increases grain yields by ~20%7,8. Hybrid rice has been planted on 55–60% (~17 million hectares) of the total rice-planting area in China, as well as in approximately 30 other countries, and thus has had a great impact on agriculture7–9. Many organellar dysfunctions,

such as mitochondrial diseases, are caused by mutations in essential organellar genes or related nuclear genes5. By contrast, CMS is usually associated with extra mitochondrial genes that can be suppressed by nuclear Rf genes, many of which encode pentatricopeptide repeat proteins2–6,10–12. Explanations proposed for CMS systems2–6,12–17 include mitochondrial energy deficiency, CMS protein cytotoxicity and premature tapetal programmed cell death (PCD). However, how CMS genes induce male sterility has not been elucidated2–6,17, and whether CMS induction involves CMS proteins interacting with nuclear-encoded mitochondrial factors is unknown17. To identify the factor(s) responsible for CMS-WA, we examined the transcripts of the whole CMS-WA mitochondrial genome by RNA blotting, and a probe containing the ribosomal protein gene rpl5 revealed a different transcript in the CMS-WA line Zhenshan 97A (ZS97A)18. Moreover, this mRNA was affected by an Rf gene. Sequence analysis of two ZS97A mitochondrial genomic clones that hybridized to rpl5 revealed a 15,742-bp rearranged DNA region comprising five segments matching rice mitochondrial and nuclear sequences and two segments of unknown origin (Supplementary Fig. 1a). This region contains a previously unidentified ORF downstream of rpl5 comprising three rice mitochondrial genomic segments and one segment of unknown origin (Fig. 1a). The 5′ region (512 bp) of this ORF is identical to that of a predicted rice mitochondrial ORF, orf284, whereas the 3′ region (583 bp) of the ORF is highly similar to that of another predicted rice mitochondrial ORF, orf288; these characteristics suggest that this recombinant structure evolved recently. This chimeric ORF encodes a 352-residue putative protein with three transmembrane segments (Supplementary Fig. 1b) and was thus named WA352 (Wild Abortive 352). Unlike most CMS genes that have similarities to known functional mitochondrial genes, such as those involved in ATP production2–4, WA352 has no similarity to these known genes. Plant mitochondrial genomic transformation is currently infeasible, but CMS gene function can be tested by nuclear transformation of candidate gene(s) fused with a mitochondrial transit signal (MTS)12,19. We constructed transformation constructs of MTSWA352, MTS-GFP-WA352 and WA352 driven by the CaMV35S

1State

Key Laboratory for Conservation and Utilization of Subtropical Agro-Bioresources, Key Laboratory of Plant Functional Genomics and Biotechnology of Guangdong Provincial Higher Education Institutions, College of Life Sciences, South China Agricultural University, Guangzhou, China. 2College of Forestry, Guangxi University, Nanning, China. 3State Key Laboratory of Biocontrol, Guangdong Key Laboratory of Plant Resources, Sun Yat-sen University, Guangzhou, China. 4Present address: College of Life Sciences, Northwest A&F University, Yangling, China. Correspondence should be addressed to Y.-G.L. ([email protected]). Received 28 September 2012; accepted 5 February 2013; published online 17 March 2013; doi:10.1038/ng.2570

Nature Genetics VOLUME 45 | NUMBER 5 | MAY 2013

573

letters

npg


Figure 1 Identification and functional analyses of the new mitochondrial gene WA352 for CMS-WA. (a) Structure of WA352, located downstream of the ribosomal protein gene rpl5 and the pseudogene ψrps14. Shown are nucleotide identities (%) of the mitochondrial genome-derived fragments to those of an indica line, 93-11, illustrating the composite origin of WA352. Long horizontal arrows indicate the transcripts. Probe 1 and probe 2 were used for RNA blotting. (b–d) Anther (top) and pollen (bottom) phenotypes of rice transgenic plants (T0 or T1 generations) with the indicated transgene constructs. The nontransgenic plant was a progeny from the cross of a heterozygous MTS-WA352 T0 plant with pollen of the host parent from the japonica variety Zhonghua11 (ZH11). The entire or truncated WA352-containing constructs were driven by the P35S promoter, and OsCOX11 RNAi was driven by the rice RTS tapetum-specific promoter26. The MTS sequence was derived from the Rf1b restorer gene12. Pollen grains were stained with 1% potassium iodide; dark staining indicates viable pollen. Scale bars, 1 mm (anthers); 50 µm (pollen).

promoter (P35S) and transferred them into the male-fertile japonica (O. sativa ssp. japonica) variety Zhonghua11 (ZH11) and Arabidopsis thaliana. Almost all (27/28) T0 transgenic plants with MTS-WA352 and most (27/42) of those with MTS-GFP-WA352 had abnormal anthers with aborted pollen but no other phenotypes, and sterility co-segregated with MTS-WA352 (Fig. 1b and Supplementary Table 1). Likewise, most (39/45) transgenic A. thaliana with MTS-WA352 were male sterile (Supplementary Fig. 2a and Supplementary Table 1). However, all 27 rice and 20 A. thaliana transgenic plants with WA352 (lacking MTS) were male fertile (Fig. 1b, Supplementary Fig. 2b and Supplementary Table 1), demonstrating that male sterility requires mitochondrial localization of WA352. Other transformations with truncated MTS-WA352 constructs and suppression of the WA352interacting gene O. sativa COX11 (OsCOX11) also produced male sterility (Fig. 1c,d and Supplementary Fig. 2b). Thus, WA352 is the causal gene for CMS-WA. We next characterized WA352 transcription. RNA blotting using probe 1 (rpl5; Fig. 1a) identified a longer transcript in ZS97A than the rpl5 transcripts in ZS97B (Fig. 2a), but probe 2 (an orf288 homolog) detected three transcripts (Fig. 2b and Supplementary Fig. 3). Quantitative RT-PCR showed that orf288, which is of unknown function, was expressed in vegetative tissues but not in anthers of a fertile line (Supplementary Fig. 4). Using circularized RNA RT-PCR for detecting transcript termini20, we identified three transcription initiation sites and two termination sites for these transcripts (Fig. 1a, Supplementary Figs. 1b and 5). The longest mRNA is dicistronic and contains rpl5 and WA352; the other two mRNAs containing WA352 only are transcribed from different initiation sites in the intergenic region, suggesting that this new gene became active through de novo transcription and co-transcription with rpl5. RNA editing is widespread for mitochondrial genes, but no editing occurred in the WA352 mRNAs. CMS-WA is genetically sporophytic21,22, meaning that WA352 acts in the diploid anther cells to cause CMS. CMS-WA can be restored by either of two dominant Rf genes, Rf3 or Rf4 (Supplementary Fig. 6), located on chromosomes 1 and 10, respectively21,22. We examined how the Rf genes affect WA352 expression using three near-isogenic lines with WA cytoplasm and the nuclear background of ZS97A but that carry different genotypes of Rf genes: ZSR1 (Rf3Rf3rf4rf4), ZSR5 (rf3rf3Rf4Rf4) and ZSR11 (Rf3Rf3Rf4Rf4)21. In the Rf4-carrying lines, the amounts of the WA352 and rpl5-WA352 transcripts were decreased to ~20–25% of those in ZS97A but were not affected by Rf3 (Fig. 2a,b and Supplementary Fig. 7a,b). This decrease may affect rpl5 function; thus, CMS-WA may be an ideal model system to study the negative 574

a

4,552 nt 1,527 nt rpl5

Probe 1

Probe 2 orf284 (100%)

orf224 (81%) Unknown origin

b

c

MTS-WA352 (T1)

MTS-WA352282–352 (T1)

2,395 nt

WA352

�rps14

Nontransgenic

MTS-WA352218–300 (T1)

orf288 (97%)

MTSGFP-WA352 (T1)

MTS-WA3521–227 (T1)

WA352 (T0)

d

OsCOX11 RNAi (T1)

pleiotropic effects and cost of fertility restoration 23. An immuno blot detected WA352 in young anthers of CMS-WA lines but not in those carrying Rf3 or Rf4 (Fig. 2c). Therefore, Rf4 and Rf3 suppress WA352 by different mechanisms, probably with Rf4 functioning posttranscriptionally and Rf3 functioning post-translationally. Notably, in CMS-WA plants, WA352 mRNAs were ubiquitously expressed (Supplementary Fig. 3), but WA352 accumulated in anthers only at the microspore mother cell (MMC) stage and not in leaves (Fig. 2c,d). In leaves of the transgenic plants, WA352 (without MTS) accumulated steadily, but the mitochondrial-targeted protein MTS-WA352 was undetectable (Fig. 2e). Furthermore, anti-GFP immunohistochemistry of the P35SøMTS-GFP-WA352 transgenic line indicated that the fusion protein was present mainly in the tapetum at the MMC stage and was largely reduced after the meiotic prophase I stage (Fig. 2f ). These results suggest that, besides suppression by the Rf genes, there are also spatial-temporal regulations of WA352 production, similarly to the male organ-specific accumulation of ORF239 in CMS lines of common bean24. Therefore, the tapetum-specific production of WA352 determines CMS specificity. To further probe the molecular mechanism of CMS-WA, we screened for rice WA352-interacting proteins by yeast two-hybrid (Y2H) assay with WA352 as bait and identified 11 candidate proteins (Supplementary Table 2). We selected a nuclear-encoded mitochondrial transmembrane protein, OsCOX11, for further analysis. COX11 proteins are conserved in eukaryotes (Supplementary Fig. 8) and function in the assembly of cytochrome c (Cyt c) oxidase25. OsCOX11 VOLUME 45 | NUMBER 5 | MAY 2013 Nature Genetics

letters

4,552

A3

52

Leaf

W

*

43 34 ZH11 (MMC)

a

WA352

352 aa

OsCOX11 TM

137


b

TM WA352- pGBKT7pGADT7 OsCOX11 OsCOX11

227

Figure 3 WA352 interacts with the nuclear-encoded mitochondrial protein COX11. (a) Interaction of WA352 with COX11 proteins of rice and A. thaliana by Y2H assay and mapping of WA352 regions (shaded) for the interaction. The combination of WA352 with the empty prey vector pGADT7, and that of the empty bait vector pGBKT7 with OsCOX11, served as negative controls. The yeast cells were grown on SD/-Leu/Trp/-His/-Ade medium. TM, predicted transmembrane segment. aa, amino acids. (b) Y2H assay mapping of the OsCOX11 region (shaded) interacting with WA352. (c) The WA352-interacting domain in OsCOX11 and the conserved sequence in A. thaliana COX11 (AtCOX11). (d) Rice protoplasts were co-transformed with four constructs that expressed, respectively, MTS-mOrange (a red fluorescent protein variant) for marking mitochondria, OsCOX11-CFP (cyan fluorescent protein) for mitochondrial localization of OsCOX11, and MTS-SPYNE-WA352218–352 and MTSSPYCE-OsCOX11 for a BiFC assay of the in vivo WA352-OsCOX11 interaction. Scale bars, 10 µm.

1

is constitutively expressed (Supplementary Fig. 9). A. thaliana AtCOX11 (287 residues), which shares 80% identity with OsCOX11, also interacted with WA352 (Fig. 3a). Y2H deletion assays identified two regions (residues 218–292 and 294–352) of WA352 that interact with OsCOX11 (Fig. 3a). Similarly, a 37-residue sequence (184–220) in the highly conserved region of OsCOX11 confers the WA352 binding (Fig. 3b,c). A bimolecular fluorescence complementation (BiFC) assay confirmed the mitochondrial localization of OsCOX11 and its in vivo interaction with WA352 (Fig. 3d and Supplementary Fig. 10). To determine whether WA352-caused sterility requires the WA352-COX11 interaction, we transferred three WA352 constructs into rice and A. thaliana; these constructs encoded truncated proteins with (MTS-WA352218–300 and MTS-WA352282–352) or without (MTS-WA3521–227) one of the COX11-binding domains (Fig. 3a). Most of the transgenic rice plants with MTS-WA352 218–300 (8/9) and MTS-WA352282–352 (9/15) were male sterile, but those with MTS-WA3521–227 were fertile (Fig. 1c and Supplementary Table 1). Transgenic Arabidopsis plants with MTS-WA352218–300 also were male sterile (Supplementary Fig. 2b and Supplementary Table 1). Furthermore, RNA interference (RNAi) of OsCOX11 driven by a tapetum-specific promoter26 also produced male sterility (Fig. 1d

and Supplementary Table 1). These results demonstrate that WA352-induced male sterility requires the COX11-interacting domains, and COX11 function is necessary for male fertility. In yeast, Saccharomyces cerevisiae COX11 (ScCOX11) has a role in hydrogen peroxide degradation27. Reactive oxygen species (ROS) affect the mitochondrial permeability transition and promote Cyt c release to the cytosol, triggering animal apoptosis or plant PCD28–30. Tapetum degeneration by PCD at the proper developmental stage(s) is crucial for pollen development16,31,32, and ROS may promote tapetal PCD33. To investigate how the WA352-COX11 interaction causes male sterility, we examined ROS in anthers using a cytochemical assay of the reaction of hydrogen peroxide with cerium chloride29. We detected high amounts of hydrogen peroxide around the mitochondrial outer membranes in the ZS97A tapetum at the MMC stage but not in ZS97B or in the later-stage tapetum of ZS97A (Fig. 4a). Immunoblotting indicated that Cyt c release occurred only in ZS97A anthers at the MMC stage (Fig. 4b,c), consistent with WA352 accumulation and the ROS burst. Furthermore, in situ TUNEL assays showed that in ZS97B and ZSR11, tapetal PCD initiated at the dyad stage for the most part; however, in ZS97A and MTS-WA352 transgenic rice, PCD occurred 1

Prophase I

X1

Prophase I

O X1

MMC

O

Sporogenous

C

npg


f

sC

MMC Meio Msp

*

55

e

W MT A3 S52

Anther (97A)

A 97

Spor

ZS

97

R 5

J2

ZS

ZS

ZS

J2

d

Leaf

3A

kDa 72

A

Anther (MMC/Meio stages)

R 1

c

3A

2,395 2,300 1,527 1,500 atp6

Figure 2 Regulation of WA352 expression. (a,b) RNA blots of WA352 transcripts in anthers with probes 1 and 2. J23A (Jin23A) and ZS97A are CMS-WA lines with recessive alleles rf3; ZS97B is the CMS maintainer line Zhenshan97B with male-fertile cytoplasm and rf3rf3 and rf4rf4; ZSR1, ZSR5 and ZSR11 are the near-isogenic lines carrying the Wild Abortive cytoplasm and nuclear background of ZS97A but also containing Rf3Rf3rf4rf4, rf3rf3Rf4Rf4 and Rf3Rf3Rf4Rf4, respectively; and Nip is the O. sativa spp. japonica line Nipponbare. Nip and ZS97B do not carry WA352 but do have orf288. The atp6 mRNA (a) and ribosomal RNAs (b) indicate equal loadings. (c,d) Immunoblot of WA352 (39 kDa; arrows) in anthers with a WA352 antibody. Spor, sporogenous cell stage; MMC/Meio, stages of microspore mother cell to meiosis I; Msp, microspore stage. (e) Transgene-expressed WA352 (arrow) detected in leaves of transgenic ZH11 plants with P35SøWA352 but not those with P35øMTS-WA352. Asterisks (c–e) indicate a cross-reacting unknown protein in all tested rice lines. (f) Immunolocalization of the fusion protein MTS-GFP-WA352 (brown) in anther sections of a male-sterile transgenic line (Fig. 1b) probed with a commercial GFP antibody. Leptoten and pachytene stages belong to meiotic prophase I. E, epidermis; En, endothecium; ML, middle layer; T, tapetum. ZH11 (MMC), nontransgenic negative control. Scale bars, 25 µm.

At

B 97

R 1 ZS R 5 ZS R 11

nt

ZS

N ip ZS 9

7A

R 11

ZSR5

ZS

97 ZS A 97 B

3A

ZS

J2

ZSR1

Probe 2

O

b

Probe 1

ZS

a

220

140

300 282

198 110

292 218

244 aa WA352

352

244 181

244

184

244

294 352

c

220 174

OsCOX11 184 AtCOX11 138

d MTS-mOrange

OsCOX11-CFP

BiFC

Merged

575

letters a

b

c

97

A

ZS

ZS

ZS97A

Bicellular pollen


Late microspore

npg

Middle microspore

f

e

Early microspore

ZS97B

d

R 11 ZS 97 B ZS 97 A ZS R 11 ZS 97 B

Figure 4 WA352 triggers a ROS burst and Cyt Cytosol Mitochondria ZS97A (MMC) ZS97B (MMC) c release to induce premature tapetal PCD and Cytosol (ZS97A) CMS. (a) Transmission electron micrograph kDa MMC Meio Msp (TEM) showing the localization of hydrogen 55 peroxide in anthers. Electron-dense deposits of 26 cerium perhydroxide (some typical deposits are ZS97A shown with arrows) indicate the accumulation 17 (metaphase I) of hydrogen peroxide around mitochondrial ACTIN1 34 outer membranes in a ZS97A tapetal cell at the COX2 26 MMC stage. No signal was detected in other samples as indicated. M, mitochondrion; N, nucleus; CW, cell wall. Scale bars, 0.5 µm. Anther MMC Prophase I Metaphase I Dyad ZS97B ZS97A stage (b,c) Immunoblot detection of rice Cyt c (12 kDa; arrows) using a commercial Cyt c antibody in cytosolic and mitochondrial protein fractions of the anthers at the MMC stage (b) and in the cytosolic fraction of different anther stages of ZS97A (c). Asterisks indicate crossreacting unknown protein(s). Meio, meiosis; Msp, microspore stage. (d) Detection of nuclear DNA fragmentation (indicating PCD) by TUNEL in developmental anthers. Red signal represents staining with propidium iodide, and yellow or RF3 RF4 green signal indicates TUNEL-positive nuclei of PCD cells. T, tapetum. Scale bars, 50 µm. Genomic Fertility WA352 WA352 restoration rearrangements (e) TEM showing tapetum degeneration in ZS97B and ZS97A; the cell breakdown in (Tapetum at Mitochondrion WA352 MMC stage) ZS97A started earlier and ended more rapidly COX11 ROS burst than that in ZS97B. En, endothecium; Cyt c Er, endoplasmic reticulum; M, mitochondrion; ROS P, plastid; Ub, ubisch body; TD, tapetum debris; RF3 RF4 COX11 Cyt c Cytosol PCW, pollen cell wall. Scale bars, 1 µm. (f) A model for the mechanism of the CMS-WA Premature tapetal PCD and degeneration Rf3 Rf4 OsCOX11 system. WA352 originated in the mitochondrial Nucleus Male sterility genome of wild rice by multiple recombination events. Rf3 and Rf4 suppress WA352 expression at different steps. In CMS-WA plants, the expression of WA352 protein is regulated to accumulate preferentially in the anther tapetum at the MMC stage. WA352 interacts with the nuclear-encoded COX11 (OsCOX11) to inhibit COX11 function in peroxide metabolism, leading to ROS burst and Cyt c release, which cause premature tapetal PCD and consequent sporophytic male sterility.

earlier, at the prophase I stage (Fig. 4d and Supplementary Fig. 11). In yeast, the knock out of ScCOX11 also caused higher amounts of ROS, Cyt c release, PCD and cell death (Supplementary Fig. 12a–d). These results indicate that COX11 proteins also function in peroxide metabolism and may act as negative regulators of PCD. To understand how the premature tapetal PCD relates to CMS-WA, we examined anther cross-sections. We found no obvious differences in anther structure and microspore generation between ZS97A and ZS97B (Supplementary Fig. 13). In ZS97B, tapetum degeneration started at the middle microspore stage (before the microspore mitosis), which was then followed by gradual deformation of organelles and loss of cytoplasm until the bicellular pollen stage (the first mitosis completed). However, in ZS97A, tapetum degeneration started at the early microspore stage and finished before the late microspore stage (Fig. 4e), which was then followed by termination of microspore development (Supplementary Fig. 13). To trace its origin, we examined WA352 in 17 wild rice species (390 accessions) and O. sativa. We detected WA352 by PCR and sequencing in some accessions of O. rufipogon (4/80), Oryza nivara (3/28) and indica rice (7/188) but not in other wild species or in japonica rice (Supplementary Table 3). Wild rice species with AA-type genomes diverged ~0.7–2 million years ago34. Therefore, WA352 probably arose 0.7–2 million years ago in an ancestor of O. rufipogon and O. nivara and was acquired by some O. rufipogon and O. nivara populations and further by indica cultivars. Indeed, we found that other types (GA, ID, Dissi, DA, K, Y and X) of rice 576

s porophytic CMS lines used for hybrid rice breeding, which were bred using female parents of indica rice or O. rufipogon9, also contained WA352 with 99.0–100% identities (Supplementary Table 3), demonstrating that they belong to the same CMS-WA type. The maintenance of WA352 in these species during evolution might be attributed to the widespread presence of Rf gene(s)35 and CMS/Rf system–based gynodioecy, a gender dimorphic genetic system with fitness advantages23. These results also suggest that indica rice had multimatrilineal origins from wild rice populations with fertile and CMS cytoplasms. Our results demonstrate that WA352, produced specifically in the tapetum at the MMC stage in CMS-WA plants, interacts with COX11 to suppress its function; this suppression induces mitochondriondriven premature tapetal PCD, thereby causing pollen abortion (Fig. 4f). This CMS model provides a mechanistic link between the gain of function of a newly identified mitochondrial CMS gene product and the loss of activity of the essential nuclear-encoded mitochondrial protein through their detrimental interaction. These findings reveal different layers of cytoplasmic-nuclear interactions for CMS induction and fertility restoration and suggest that plants have evolved complex systems for counteracting the negative effects of new mitochondrial genes resulting from genomic rearrangements. Some other CMS systems, such as sunflower CMS-PET1, which expresses the CMS-associated protein ORF522 (ref. 14) and undergoes premature tapetal PCD16, may also involve similar mechanisms of mitochondrial-nuclear gene interactions. This study provides insights into the evolutionary importance and biochemical mechanisms of VOLUME 45 | NUMBER 5 | MAY 2013 Nature Genetics

letters cytoplasmic-nuclear genomic interactions and incompatibilities and has practical implications for hybrid crop breeding. Methods Methods and any associated references are available in the online version of the paper. Accession codes. The WA352 gene and protein sequences are deposited in GenBank under accession number JX131325. Note: Supplementary information is available in the online version of the paper.

npg


Acknowledgments We thank G. Zhang, X. Liu (South China Agricultural University) and Q. Qian (China National Rice Research Institute) for providing some rice materials, H. Pang for technical assistance (GIBH, CAS) and H. Ma (Fudan University), J.-M. Li (University of Michigan) and Z.-H. He (San Francisco State University) for critical reading of the manuscript. This work was supported by grants (2011CB100203/2013CBA01401, 31230052 and 2013CB126900) from the Ministry of Science and Technology of China and the National Natural Science Foundation of China to Y.-G.L. AUTHOR CONTRIBUTIONS D.L. performed most of the experiments involving the CMS gene functions and mechanism. H.X. and Z.L. found the CMS gene and analyzed its expression. J.G. investigated the gene origin. H.L., L.C., C.F., M.B., N.Y. and Hong Wu conducted cytological observations. Hao Wu and C.J. did plant transformation. Q.Z. mapped the mRNA ends. H.Z. mapped the interaction domains. Y.C., S.Y., X.L., X.Z. and R.L. participated in the characterization of the genes. Y.-G.L. designed and supervised the study and wrote the manuscript. All of the authors discussed the results and commented on the manuscript. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html. 1. Laser, K. & Lersten, N. Anatomy and cytology of microsporogenesis in cytoplasmic male sterile angiosperms. Bot. Rev. 38, 425–454 (1972). 2. Schnable, P.S. & Wise, R.P. The molecular basis of cytoplasmic male sterility and fertility restoration. Trends Plant Sci. 3, 175–180 (1998). 3. Hanson, M.R. & Bentolila, S. Interactions of mitochondrial and nuclear genes that affect male gametophyte development. Plant Cell 16, S154–S169 (2004). 4. Chase, C.D. Cytoplasmic male sterility: a window to the world of plant mitochondrialnuclear interactions. Trends Genet. 23, 81–90 (2007). 5. Woodson, J.D. & Chory, J. Coordination of gene expression between organellar and nuclear genomes. Nat. Rev. Genet. 9, 383–395 (2008). 6. Carlsson, J. & Glimelius, K. Cytoplasmic male-sterility and nuclear encoded fertility restoration. in Plant Mitochondria (ed. Kempken, F.) 469–491 (Springer, New York, 2011). 7. Lin, S. & Yuan, L. Hybrid rice breeding in China. in Innovative Approaches to Rice Breeding 35–51 (IRRI, Manila, 1980). 8. Virmani, S.S. Advances in hybrid rice research and development in the tropics. in Proceedings of the 4th International Symposium on Hybrid Rice (eds. Virmani, S.S., Mao, C.X. & Hardy, B.) 7–20 (IRRI, Manila, 2003). 9. Cheng, S.H., Zhuang, J.Y., Fan, Y.Y., Du, J.H. & Cao, L.Y. Progress in research and development on hybrid rice: a super-domesticate in China. Ann. Bot. (Lond.) 100, 959–966 (2007). 10. Bentolila, S., Alfonso, A.A. & Hanson, M.R. A pentatricopeptide repeat-containing gene restores fertility to cytoplasmic male-sterile plants. Proc. Natl. Acad. Sci. USA 99, 10887–10892 (2002).


11. Brown, G.G. et al. The radish Rfo restorer gene of Ogura cytoplasmic male sterility encodes a protein with multiple pentatricopeptide repeats. Plant J. 35, 262–272 (2003). 12. Wang, Z. et al. Cytoplasmic male sterility of rice with boro II cytoplasm is caused by a cytotoxic peptide and is restored by two related PPR motif genes via distinct modes of mRNA silencing. Plant Cell 18, 676–687 (2006). 13. Warmke, H.E. & Lee, S.L. Pollen abortion in T cytoplasmic male-sterile corn (Zea mays): a suggested mechanism. Science 200, 561–563 (1978). 14. Sabar, M., Gagliardi, D., Balk, J. & Leaver, C.J. ORFB is a subunit of F1F(O)-ATP synthase: insight into the basis of cytoplasmic male sterility in sunflower. EMBO Rep. 4, 381–386 (2003). 15. Levings, C.S. III. Thoughts on cytoplasmic male sterility in cms-T maize. Plant Cell 5, 1285–1290 (1993). 16. Balk, J. & Leaver, C.J. The PET1-CMS mitochondrial mutation in sunflower is associated with premature programmed cell death and cytochrome c release. Plant Cell 13, 1803–1818 (2001). 17. Guo, J. & Liu, Y.-G. The genetic and molecular basis of cytoplasmic male sterility and fertility restoration in rice. Chin. Sci. Bull. 54, 2404–2409 (2009). 18. Liu, Z., Xu, H., Guo, J. & Liu, Y.-G. Structural and expressional variations of the mitochondrial genome conferring the Wild Abortive type of cytoplasmic male sterility in rice. J. Integr. Plant Biol. 49, 908–914 (2007). 19. He, S., Abad, A.R., Gelvin, S.B. & Mackenzie, S.A. A cytoplasmic male sterilityassociated mitochondrial protein causes pollen disruption in transgenic tobacco. Proc. Natl. Acad. Sci. USA 93, 11763–11768 (1996). 20. Zhang, Q. & Liu, Y.-G. Rice mitochondrial genes are transcribed by multiple promoters that are highly diverged. J. Integr. Plant Biol. 48, 1473–1477 (2006). 21. Zhang, G., Lu, Y., Bharaj, T.S., Virmani, S.S. & Huang, N. Mapping of the Rf-3 nuclear fertility-restoring gene for WA cytoplasmic male sterility in rice using RAPD and RFLP markers. Theor. Appl. Genet. 94, 27–33 (1997). 22. Zhang, Q.Y., Liu, Y.-G., Zhang, G. & Mei, M. Molecular mapping of the fertility restorer gene Rf-4 for WA cytoplasmic male sterility in rice. Yi Chuan Xue Bao 29, 1001–1004 (2002). 23. Delph, L.F., Touzet, P. & Bailey, M.F. Merging theory and mechanism in studies of gynodioecy. Trends Ecol. Evol. 22, 17–24 (2007). 24. Abad, A.R., Mehrtens, B.J. & Mackenzie, S.A. Specific expression in reproductive tissues and fate of a mitochondrial sterility-associated protein in cytoplasmic malesterile bean. Plant Cell 7, 271–285 (1995). 25. Banting, G.S. & Glerum, D.M. Mutational analysis of the Saccharomyces cerevisiae cytochrome c oxidase assembly protein Cox11p. Eukaryot. Cell 5, 568–578 (2006). 26. Luo, H. et al. RTS, a rice anther-specific gene is required for male fertility and its promoter sequence directs tissue-specific gene expression in different plant species. Plant Mol. Biol. 62, 397–408 (2006). 27. Veniamin, S., Sawatzky, L.G., Banting, G.S. & Glerum, D.M. Characterization of the peroxide sensitivity of COX-deficient yeast strains reveals unexpected relationships between COX assembly proteins. Free Radic. Biol. Med. 51, 1589–1600 (2011). 28. Liu, X., Kim, C.N., Yang, J., Jemmerson, R. & Wang, X. Induction of apoptotic program in cell-free extracts: requirement for dATP and cytochrome c. Cell 86, 147–157 (1996). 29. Yao, N. et al. Mitochondrial oxidative burst involved in apoptotic response in oats. Plant J. 30, 567–579 (2002). 30. Greenberg, J.T. & Yao, N. The role and regulation of programmed cell death in plant–pathogen interactions. Cell Microbiol. 6, 201–211 (2004). 31. Diamond, M. & McCabe, P.F. Mitochondrial regulation of plant programmed cell death. in Plant Mitochondria: Advances in Plant Biology 1 (ed. Kempken, F.) 439–465 (Springer Science Business Media, 2011). 32. Papini, A., Mosti, S. & Brighigna, L. Programmed-cell-death events during tapetum development of angiosperms. Protoplasma 207, 213–221 (1999). 33. Hu, L. et al. Rice MADS3 regulates ROS homeostasis during late anther development. Plant Cell 23, 515–533 (2011). 34. Zhu, Q. & Ge, S. Phylogenetic relationships among A-genome species of the genus Oryza revealed by intron sequences of four nuclear genes. New Phytol. 167, 249–265 (2005). 35. Li, S. et al. Distribution of fertility-restorer genes for Wild-Abortive and Honglian CMS lines of rice in the AA genome species of genus Oryza. Ann. Bot. (Lond.) 96, 461–466 (2005).

577

ONLINE METHODS

npg


Cloning of the recombinant region of the CMS-WA mitochondrial genome. The mitochondrial genomic DNA of ZS97A was partially digested with HindIII and separated on a low-melting agarose gel. DNA fragments of ~20–30 kb were recovered and cloned into the TAC vector36. Two overlapped positive clones with insert sizes of 29 kb and 30.7 kb were identified by colony hybridization with the rpl5 probe and sequenced. Preparation of binary constructs and plant transformation. The DNA fragments for expressing the full-length or truncated WA352 proteins (WA352, WA3521–227, WA352218–300 and WA352282–352) were amplified by PCR using the appropriate primers (all primers are listed in Supplementary Table 4) and used to replace the orf79 sequence of the binary construct P35SøMTS-orf79 (ref. 12), where the MTS sequence (1–105 bp) was derived from Rf1b of the rice CMS-BT system12. A GFP sequence amplified with forward and reverse primers (GFP-F and GFP-R) was inserted in the PstI site linking MTS and WA352 in the P35SøMTS-WA352 construct to produce P35SøMTS-GFP-WA352. To generate the control WA352 construct without MTS, the WA352 sequence was amplified with the primers WA352-24F and WA352R and replaced the MTSWA352 sequence in P35SøMTS-WA352 to produce P35SøWA352. To prepare the OsCOX11 RNAi construct, the promoter sequence of the rice RTS gene26 was amplified by PCR using primers RTS-F and RTS-R and cloned into the pYLRNAi vector37 to replace the maize ubiquitin 1 promoter. The DNA fragment of OsCOX11 (388–708 bp) was amplified using the primer pairs OsCOX11-388F and OsCOX11-708R and used to generate the RNAi constructs as described38. The binary constructs were transferred into the rice variety ZH11 and A. thaliana (ecotype Columbia) by Agrobacterium tumefaciens–mediated transformation39,40. The in vitro germination of Arabidopsis pollen was done as described41. Expression analyses. Total RNA was extracted from rice tissues using TRIzol Reagent (Invitrogen). For RNA blotting, probes were labeled with 32P-dCTP using a random-primed labeling kit (Takara, Dalian China). The expression of OsCOX11 was assayed by semiquantitative RT-transcription PCR. Y2H assay. WA352 fragments of various lengths were amplified using the WA352 primers for Y2H assay and cloned into the bait vector pGBKT7 (Clontech, USA). The COX11 fragments from rice and A. thaliana were amplified using the COX11 primers and cloned into the prey vector pGADT7. The bait and prey constructs were co-transformed into yeast strain AH109, and the transformed cells were plated on synthetic defined medium containing all essential amino acids except leucine and tryptophan (SD/-Leu/-Trp) medium and incubated at 30 °C for 3 d and then diluted and applied onto SD/-Leu/Trp/-His/-Ade medium and SD/-Leu/-Trp medium (as loading control) and cultured at 30 °C for 2-3 days. Subcellular localization of OsCOX11 and BiFC assay. The OsCOX11 sequence was amplified using the primers s-OsCOX11F and s-OsCOX11R and cloned into a pUC18-based CFP gene vector driven by the P35S promoter. The MTS sequence12 was amplified using the primers Rf1b-1F and Rf1b-35R and cloned into a mOrange-containing vector42. To prepare the BiFC constructs, the MTS sequence was cloned into the BiFC vectors pUC-SPYNE and pUCSPYCE37 to produce pUC-MTS-SPYNE and pUC-MTS-SPYCE, respectively. Fragments of WA352218–352 and OsCOX11 were amplified using the primer pairs b-WA352-218F and WA352R and b-OsCOX11F and b-OsCOX11R, respectively, and cloned into the modified BiFC vectors to produce pUC-MTSSPYNE-WA352218–352 and pUC-MTS-SPYCE-OsCOX11 (note, OsCOX11 was linked downstream of SPYCE, and its own MTS sequence was no longer located at the 5′ region of the recombinant gene; therefore an additional MTS sequence was added to the 5′ region of the construct). Transformation of protoplasts was performed as described42. After incubation of the transformed protoplasts at 30 °C for 24 h, fluorescence images for mOrange, CFP and yellow fluorescent protein (YFP) were obtained using a laser scanning confocal microscope (Leica-TCS SP2) with excitation and emission spectrum analysis (Supplementary Fig. 9).

Nature Genetics

Preparation of mitochondrial and cytosolic extracts. Etiolated 10-dayold rice seedlings (7–8 g) were ground in 30 ml of homogenization solution (500 mM sucrose, 100 mM Tris-HCl, 100 mM ethylenediaminetetraacetic acid (EDTA)-Na2, 0.1% BSA and 0.5% polyvinylpyrrolidone-40, pH 7.5). After filtering the homogenate through four layers of Miracloth, the mixture was sequentially centrifuged at 1,000g for 5 min, 3,000g for 5 min, 6,000g for 5 min and 12,000g for 15 min. The pellets with enriched mitochondria were suspended in 15 ml wash buffer (300 mM sucrose, 50 mM Tris-HCl, 0.1% BSA and 0.5% polyvinylpyrrolidone-40, pH 7.5) and centrifuged at 1,000g for 5 min to remove any tissue debris, and then the supernatants were further centrifuged at 12,000g for 20 min to pellet the mitochondria. The primary supernatants were centrifuged at 20,000g for 1 h to remove residual mitochondria. The supernatants recovered from this step contained cytosolic proteins. The mitochondrial and cytosolic proteins were stored at –70 °C. The anther mitochondria were isolated from 1 g of anthers using the same procedures, except that the tissues were ground in 3 ml homogenization solution, and the filtration step was omitted. Immunoblot analysis. A synthetic polypeptide (RSPPYAPYPYPVDE) corresponding to residues 224–237 of WA352, which was selected on the basis of an antigenic epitope analysis, was used to immunize rabbits to obtain the antibody serum by AbMART (Shanghai, China). Mitochondrial pellets (50 µg) were lysed in 50 µl lysis solution (100 mM Tris-HCl, 5 mM EDTA, 5 mM dithiothreitol (DTT), 1% SDS and 0.1% Triton X-100, pH 8.0). After centrifugation at 12,000g for 15 min, the supernatants were mixed with an equal volume of 2× loading buffer (100 mM Tris-HCl, 5 mM DTT, 4% SDS, 0.01% bromophenol blue and 30% glycerol, pH 6.8). Cytosolic protein samples were mixed with an equal volume of the 2× loading buffer, separated by 12% SDS-PAGE and transferred onto an Immobilon-PPSQP transfer membrane (polyvinylidene fluoride (PVDF) type, Millipore) using a Bio-Rad mini transfer cell. The membrane blots were incubated in blocking buffer (5% milk, 0.1% Tween-20, 0.1% Triton X-100, 100 mM Tris-HCl and 150 mM NaCl, pH 7.5) for 2 h at room temperature, washed twice with TBSTT buffer (0.1% Tween-20, 0.1% Triton X-100, 100 mM Tris-HCl and 150 mM NaCl, pH 7.5) and incubated with WA352 antibody (1:10,000 dilution) for 20 h at 4 °C. After two rinses with TBSTT, the blots were incubated in 1:10,000diluted secondary antibody solution (affinity-purified HRP-conjugated Affinipure goat anti-rabbit IgG (H+L), cat. no. SA00001-2, ProteinTech Group Inc.) for 1 h at room temperature and washed twice with TBSTT. The membrane blots were incubated in the ECL substrate (Advansta, USA) for 2 min. To detect Cyt c, the blots were probed with a 1:2,000-diluted monoclonal antibody (clone 7H8.2C12, BD Biosciences). To detect COX2, the blots were probed with a polyclonal antibody (art no. AS04 053A, Agrisera) in a 1: 2,000 dilution. Immunolocalization of GFP-WA352. As the WA352 antibody cross-reacted with unknown proteins (Fig. 2c–e), we used the male-sterile transgenic lines with P35SøMTS-GFP-WA352 (Supplementary Table 1) for an immuno histochemistry assay with a GFP antibody (code no. M048-3, MBL, USA). The anthers were fixed in 4% paraformaldehyde solution, embedded in paraffin and sectioned at a thickness of 6 µm. The sections were deparaffinized, dehydrated and then blocked to remove endogenous peroxidase with 3% hydrogen peroxide for 10 min and heated with microwaves in 0.01 M PBS buffer (pH 7.2) for 5 min to retrieve antigens. The sections were immunostained by the streptavidin-biotin complex (SABC) method (Boster Company, China). The sections were incubated with 5% BSA for 20 min at room temperature to eliminate nonspecific staining and then incubated with the GFP antibody (dilution 1:100) for 3 h at 20 °C. After washing with PBS, the sections were incubated with the secondary antibody (cat. no. SA1021, Boster, China) in a 1:100 dilution for 30 min, followed by reaction with avidin-biotin-peroxidase for 20 min at room temperature. The slides were visualized with DAB (3,3′-diaminobenzidine) reagent. Cytochemical localization of hydrogen peroxide in anthers. To detect hydrogen peroxide localization in the anther cells, the histochemical cerium chloride method based on the generation of cerium perhydroxide was performed as described29. Rice anthers at various developmental stages were incubated in 10 mM CeCl3 (Sigma-Aldrich, USA) in 50 mM 3-(N-morpholino) propanesulfonic

doi:10.1038/ng.2570

acid (MOPS) buffer at pH 7.2 for 80 min. The samples were prefixed in 2.5% (vol/vol) glutaraldehyde and 2% (vol/vol) paraformaldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) overnight at 4 °C. After three washes with cacodylate buffer, the samples were post-fixed with 1% (wt/vol) osmium tetroxide in cacodylate buffer for 1 h, dehydrated in an ethanol series and then embedded in Eponate 12 resin (Ted Pella Inc., Redding, CA). Ultrathin sections (100 nm) were obtained on a Microtome (Leica EM UC6, Vienna, Austria), mounted on nickel grids (200 mesh) and examined without staining with a transmission electron microscope (JEM-1400, JEOL, Tokyo, Japan). TUNEL assay for anther PCD. The anther developmental stages were confirmed by observation of cross-sections of anthers from the same flowers by light microscopy. The preparation of the anther sections and the TUNEL assay with a Dead End Fluorometric TUNEL Kit (Promega) were performed as described43. The green fluorescence of fluorescein (TUNEL signal) and red fluorescence of propidium iodide were analyzed at 488 nm (excitation) and 520 nm (detection), and 488 nm (excitation) and 610 nm (detection), respectively, under a laser scanning confocal microscope (Leica-TCS SP2).

36. Liu, Y.-G. Complementation of plant mutants with large genomic DNA fragments by a transformation-competent artificial chromosome vector accelerates positional cloning. Proc. Natl. Acad. Sci. USA 96, 6535–6540 (1999). 37. Walter, M. et al. Visualization of protein interactions in living plant cells using bimolecular fluorescence complementation. Plant J. 40, 428–438 (2004). 38. Hu, X. & Liu, Y.-G. The construction of RNAi vectors and the use for gene silencing in rice. Mol. Plant Breed. 4, 621–626 (2006). 39. Hiei, Y., Ohta, S., Komari, T. & Kumashiro, T. Efficient transformation of rice (Oryza sativa L.) mediated by Agrobacterium and sequence analysis of the boundaries of the T-DNA. Plant J. 6, 271–282 (1994). 40. Clough, S.J. & Bent, A.F. Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana. Plant J. 16, 735–743 (1998). 41. Fan, L.M., Wang, Y.F., Wang, H. & Wu, W.H. In vitro Arabidopsis pollen germination and characterization of the inward potassium currents in Arabidopsis pollen grain protoplasts. J. Exp. Bot. 52, 1603–1614 (2001). 42. Chen, L. et al. The Hop/Sti1-Hsp90 chaperone complex facilitates the maturation and transport of a PAMP receptor in rice innate immunity. Cell Host Microbe 7, 185–196 (2010). 43. Li, N. et al. The rice tapetum degeneration retardation gene is required for tapetum degradation and anther development. Plant Cell 18, 2999–3014 (2006).

npg


Light microscopy and electron microscopy of anthers. Preparation of sections of rice anthers for light microscopy and electron microscopy were

performed as described43. The sections were examined and photographed using a Phillips FEI-TECHNAI 12 TEM (Eindhoven, the Netherlands).

doi:10.1038/ng.2570

Nature Genetics