Cytoplasmic male sterility (CMS) has attracted great interest because of its application in crop breeding. Despite increasing knowledge of CMS, not much is understood about its molecular mechanisms. Previously, orfH79 was cloned and identified as the CMS gene in Honglian rice, but how the ORFH79 protein causes pollen abortion is still unknown.
Through bacterial two-hybrid library screening, P61, a subunit of the mitochondrial electron transport chain (ETC) complex III, was selected as a candidate that interacts with ORFH79. Bimolecular fluorescence complementation (BiFC) and coimmunoprecipitation (coIP) assays verified their interaction inside mitochondria. Blue native polyacrylamide gel electrophoresis (BN-PAGE) and western blotting showed ORF79 and P61 colocalized in mitochondrial ETC complex III of CMS lines.
Compared with the maintainer line, Yuetai B (YB), a significant decrease of enzyme activity was detected in mitochondrial complex III of the CMS line, Yuetai A (YA), which resulted in decreased ATP concentrations and an increase in the reactive oxygen species (ROS) content.
We propose that the CMS protein, ORFH79, can bind to complex III and decrease its enzyme activity through interaction with P61. This defect results in energy production dysfunction and oxidative stress in mitochondria, which may work as retrograde signals that lead to abnormal pollen development.
Cytoplasmic male sterility (CMS), a maternally inherited phenomenon existing widely in the plant kingdom (Lewis, 1941; Laser & Lersten, 1972; Janska & Mackenzie, 1993; Krishnasamy & Makaroff, 1994; Hanson & Bentolila, 2004), is an evolutionarily subtle means for plants to acquire foreign pollen to increase genetic diversity by preventing self-pollination (Dewey et al., 1986; Young & Hanson, 1987; Hanson, 1991; Smart et al., 1994). CMS has been a useful approach employed for heterosis in plant breeding programs.
Many studies suggest that the CMS-associated genes are always chimeric and cotranscribed with genes encoding the mitochondrial subunits of the electronic transport chain (ETC) or ATP synthase complex (Chase, 2007; Kubo & Newton, 2008; Arrieta-Montiel & Mackenzie, 2011) or truncated versions of the functional mitochondrial genes for ETC or ATP synthase complex. Based on these facts, two hypotheses have been proposed for the molecular mechanism of CMS (Budar et al., 2003; Chase, 2007).
The first hypothesis is known as the gain-of-function model. It emphasizes the potential involvement of the CMS gene in an unknown process that hurts the mitochondria. There are at least two examples that support this model. One is the Texas male sterile cytoplasm (T-cms) maize protein, T-URF13, which was thought to form a pore in the inner mitochondrial membrane to disrupt the membrane potential (Rhoads et al., 1995; Wise et al., 1998). Another example is the recently identified sterility-inducing protein, ORF138, in Ogura rapeseed. Similar to T-URF13 in maize, this protein is involved in an unknown complex mainly composed of itself that forms a pore in the inner mitochondrial membrane (Duroc et al., 2009).
The second model is termed loss-of-function, and is supported by more cases. Connett & Hanson (1990) first reported that different mitochondrial ETC complexes were observed between male sterile and fertile Petunia. A study of wild G CMS beets found instability of complex IV and a decrease of cytochrome c oxidase activity in mitochondria (Ducos et al., 2001). In sunflower, the CMS protein ORF522 has sequence identity with a subunit of ATP synthase. The authors speculated that ORF522 can compete with the normal subunit to impair the function of the ATP synthase complex (Sabar et al., 2003). All of these studies point to the association among the CMS protein and the ETC complexes of mitochondria. Thus, we speculate that the CMS proteins somehow bind to some mitochondrial proteins to achieve such an effect. However, to date, no candidate protein has been identified that directly interacts with CMS proteins.
Rice of Honglian (HL) CMS (HL-CMS), a gametophytic CMS derived from common wild rice (Oryza rufipogon), has been the main type of CMS used for hybrid rice production in China and southern Asia (Tan et al., 2012). The area cultivated with the HL-type hybrid has cumulatively reached almost 10 million hectares during the past several decades (Li et al., 2007). The pollen of HL-CMS halts development at the bicellular stage and exhibits spherical abortion (Supporting Information, Fig. S1). We have previously found that the chimeric gene, orfH79, is responsible for the male sterility in HL-CMS (Yi et al., 2002). A large accumulation of ORFH79 in mitochondria occurs during pollen development in HL-CMS lines (Hu et al., 2012), which consequently causes an increase in reactive oxygen species (ROS) and decrease in the ATP/ADP ratio in the anthers (Li et al., 2004; Wan et al., 2007; Peng et al., 2010). This suggests that severe energy deficiency in these mitochondria is possibly the biochemical cause of CMS, while the detailed mechanism of pollen disruption remains unclear. In this study, a subunit of mitochondria complex III, p61, was identified via bacterial two-hybrid library screening as a candidate protein that directly interacts with ORFH79. We further verified this interaction with other in vivo and in vitro assays. Our results shed much light on how the plant CMS genes function at the molecular level.
Materials and Methods
The mitochondria were purified from etiolated rice (Oryza sativa L.) seedlings of the HL-CMS line, Yuetai A (YA), and the corresponding maintainer, Yuetai B (YB), by differential and density gradient centrifugation (Eubel et al., 2007). Soluble and membrane-bound (insoluble) mitochondrial proteins were prepared as previously reported (Uyttewaal et al., 2008).
Construction and screening of bacterial two-hybrid library
Following the instructions of the BacterioMatch® II Two-Hybrid System Library Construction Kit (Stratagene, La Jolla, CA, USA), the cDNA library from young inflorescences of F1 HL rice was used to construct the two-hybrid target plasmid, pTRG. The CMS gene, orfH79, was cloned into the bait pBT vector between the BamHI and XholI sites. Self-activation test and library screening and validation were also performed as the instructions described. The candidate positive clones were sequenced and annotated by Web BLAST Search of the Rice Genome Annotation Project (http://rice.plantbiology.msu.edu/).
Bimolecular fluorescence complementation (BiFC) assay in rice protoplasts
The transit peptide of atpγ fused to the N-terminus of orfH79 was synthesized by GenScript® (Nanjing, China). The fusion gene Gama-H79, the transit peptide of atpγ, and orfH79 were inserted into pUC-SPYCE (C-terminal of YFP) via BamHI and XholI. The P61 was inserted into pUC-SPYNE (N-terminal) via XholI and KpnI. All constructs were fully sequenced to ascertain correct insertion.
For transient expression assays, polyethylene glycol (PEG)-mediated transformation was performed using rice protoplasts that were prepared from rice seedlings (7–10 d; Bart et al., 2006). The cotransformed vector, FA-RFP, encoding ATPase-γ (a subunit of ATPase complex F1) fused with red fluorescent protein (RFP), was used as a mitochondrial marker. The fluorescent signals and bright-field images were taken by a FV1000 confocal system (Olympus, Tokyo, Japan).
Lysis of mitochondria
For coIP, 300 μg of each mitochondria (YA &YB) were lysed by gentle agitation for 20 min at 4°C in 100 μl IP (immunoprecipitation) buffer (10 mM HEPES/KOH, 50 mM NaCl, 2 mM EDTA (pH 7.4), Complete™ EDTA-free protease inhibitor cocktail (Roche Applied Science) with 1% (w/v) digitonin. Then, 1.9 ml IP buffer with 0.1% (w/v) digitonin was added. Both of the extracts were isolated by centrifugation for 30 min at 20 000 g at 4°C. The supernatants were used for the next round of coIP experiments.
Coupling of antibodies to protein A-sepharose beads
The C-terminal polypeptide (DWIKKTFFEKPEPE) of P61 was used to prepare the polyclonal antibody in rabbit (BPI, Beijing, China). The ORFH79 rabbit antibody was previously produced by our laboratory (Peng et al., 2010). Before coupling, 100 μl of protein A beads were washed three times in IP buffer (0.1% digitonin) and incubated for 40 min at room temperature (20–25°C) with either 10 μg antibodies against ORFH79 or 10 μg antibodies against P61. The beads were again washed three times with IP buffer (0.1% digitonin).
The YA extract (1 ml each) was added separately to the beads coupled with antibody against ORFH79 or P61. The YB extract was treated the same as the control. After overnight incubation at 4°C, the beads were washed three times with 1 ml IP buffer (0.1% digitonin), resuspended in 50 μl 2 × sodium dodecyl sulfate (SDS) gel loading buffer, and incubated for 3 min at 96°C. Isolated proteins were separated by 10% Tricine-SDS-polyacrylamide gel electrophoresis (PAGE) and analyzed by western blot using antisera against P61 and ORFH79, separately.
Blue native PAGE
Mitochondrial extracts (0.4 mg protein each of YA and YB) were resuspended in 47.5 μl 0.75 M aminocaproic acid, 50 mM Bis–Tris (pH 7.0; 1% n-dodecyl-β-d-maltopyranoside (DDM)). After incubating for 30 min on ice, the mixture was centrifuged for 30 min at 24 000 g. The supernatant was collected, then 2.5 μl 5% Coomassie blue G in 0.5 M aminocaproic acid and Complete™ EDTA-free protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany) were added. A gradient gel of 5–13% acrylamide was subsequently used to resolve protein complexes. To verify P61 and ORFH79 location, the band representing complex III was cut to run the second dimension (10% Tricine-SDS-PAGE) and analyzed by western blotting.
Far western blotting
The N-terminal (1–24 aa) and C-terminal (41–61 aa) peptides of P61 were synthesized by GenScript®. The orfH79 gene fused with glutathione S-transferase (GST) protein and GST protein alone were expressed and purified by Escherichiacoli BL21 with pGEX5x-2 (GE Healthcare, Piscataway, NJ, USA). The orfH79 gene was inserted into the BamHI and XholI sites of pGEX5x-2. Far western blotting was performed as described previously (Wheeler et al., 2009), with minor modifications. Nitrocellulose membranes were spotted with 5 μg of N24 and C21 peptides of P61. The membranes were blocked by incubation in PBST buffer (100 mM phosphate, pH 7.5, 100 mM NaCI, 0.1% Tween 20) with 5% nonfat milk, for 2 h at room temperature. The membranes were then incubated overnight at 4°C with PBST (5% nonfat milk) with either 4 mg ml−1 orfH79-GST or GST. The next day, the membranes were washed and incubated with rabbit anti-GST antibody (1 : 5000 dilution) for 2 h in PBST (5% nonfat milk) at room temperature. Lastly, the membranes were incubated in horseradish peroxidase-linked secondary antibody (1 : 10 000), and ECL Plus reagent (Thermo Scientific) was used for exposure to X-ray film.
Mitochondrial enzyme activity assays
Before testing enzyme activity, YA and YB mitochondria lysates were prepared as done for the coIP assays and quantified using a BCA Protein Quantitative Reagent Kit (Thermo Scientific). At the start of the activity test, 10 μg of mitochondria per reaction were used. Ubiquinol cytochrome c reductase (complex III) activity and ATP synthase (complex V) activity were measured at 25°C, according to the instructions of the Complex III Activity kit, the Complex V Activity Kit (Genmed, Shanghai, China), and a previously described protocol (Kirby et al., 2007). Absorbances at 550 nm (for complex III) and 340 nm (for complex V) were measured every 20 s for 5 min on a multifunctional microplate reader (SpectraMax M2; Molecular Devices, Sunnyvale, CA, USA) in kinetic mode. The data were analyzed by SoftMax Pro (Molecular Devices). To test the impact of ORFH79 on normal mitochondria in vitro, lysed YB mitochondria were incubated with 30 μg ml−1 of ORFH79-GST fusion protein for 30 min. Incubation with 30 and 300 μg ml−1 GST protein for 30 min served as controls.
Quantification of ATP and ROS
The ATP assay kit (Beyotime, Jiangsu, China), which employs the luciferin-luciferase method (Drew & Leeuwenburgh, 2003), was used to quantify ATP. The working solution was prepared according to the kit protocol. Then, 100 μl of working solution and 100 μl of purified mitochondria (100 μg ml−1) were added to each well of a 96-well microtiter plate. The luciferase signals was detected by a multifunctional microplate reader (SpectraMax M2) for 30 s. The standard curve of ATP concentration from 1 pM to 1 μM was prepared by gradient dilution.
For ROS quantification, the Reactive Oxygen Species Assay Kit (Beyotime) was used. To each well of a 96-well microtiter plate, 100 μl of 20 μM dichlorofluorescin diacetate (DCFDA; Molecular Probes, OR, USA) in phosphate-buffered saline with dissolved dimethyl sulfoxide and 100 μl purified mitochondria (100 μg ml−1) were added. The fluorescent DCF signals, which resulted when DCFDA was oxidized by H2O2 and other peroxides, were also detected with a multifunctional microplate reader (SpectraMax M2) by monitoring emission at 520 nm with an excitation wavelength of 485 nm.
Real-time PCR and western blotting
The rice homologs of CDKA;1, KRP6/7, RHF, and FBL17a in Arabidopsis were analyzed using the GRAMENE website (http://www.gramene.org/cmap/). Anthers of YB at unicellular, bicellular, and mature stages and the corresponding length of YA anthers were acquired and ground using liquid nitrogen. The RNAs were prepared using TRIzol and proteins were then precipitated from the phenol-ethanol supernatant with isopropyl alcohol according to the instructions of the TRIzol kit (Invitrogen, Carlsbad, CA, USA). Reverse transcription and real-time PCR (SYBRO Green I-based method) were performed using the Lightcycler 480 system (Roche Diagnostics, Mannheim, Germany) as previously described (Wan et al., 2007). For real-time PCR, actin was used as internal control and the sequences of the primers are listed in Table S3. For semiquantitative PCR, 18S rRNA was used as control and 25 cycles was performed; the sequences of the primers are listed in Table S4. Western blotting was performed as previously described (Peng et al., 2010). The antiserum for the KRP6/7 rice homolog was prepared by polypeptide (ERRRFAEKYNYDIALDRPLQGRC) immunization of rabbit (Neweastbio, Wuhan, China). The antibodies for CDKA1 and phosphorylated CDKA1 were ordered from Cell Signaling Technology (Beverly, MA, USA).
Bacterial two-hybrid system screening
Based on the loss-of-function model, there should be a functional protein that can interact with CMS proteins in mitochondria. In order to search for the interacting partners of ORFH79, bacterial two-hybrid screening of the young rice inflorescence cDNA library (1.03 × 106 capacity) was conducted (Wang et al., 2009). The pBT-orfH79 vector was used as the bait vector to screen the library. Several clones were selected from the final dual selective medium plate. The sequencing results identified 12 candidate clones, of which a novel candidate attracted our interest upon subsequent bioinformatic analysis. This protein was 6.7 kDa and annotated as a putative subunit of the ubiquinol-cytochrome c reductase complex according to GRAMENE (http://www.gramene.org/). Our screening and validation results are shown in Fig. 1. The full-length cDNA of P61 (Loc_Os07 g48244) contains a 186 base pair open reading frame encoding a protein with 61 amino acid residues. Thus, we named it P61.
P61 is a subunit of complex III in mitochondria
Previous studies showed that ORFH79 was localized in mitochondria of CMS line YA (Peng et al., 2010). Thus, P61 localization in mitochondria was the precondition for its interaction with ORFH79 in vivo. To determine if P61 is localized in mitochondria, a bioinformatic prediction was performed first. The online subcellular prediction software, IPSORT (http://ipsort.hgc.jp/), Mito Prot II (http://ihg.gsf.de/ihg/mitoprot.html), and TargetP (http://www.cbs.dtu.dk/services/TargetP/), all predicted that P61 was localized in mitochondria (Table S1). To verify its localization in mitochondria, western blot analysis of total proteins extracted from etiolated seedlings, purified mitochondria, and insoluble and soluble fractions of mitochondria was carried out. We observed that the band from purified mitochondria (lane 2, Fig. 2a) was stronger than that from the total cell proteins of etiolated seedlings (lane 1, Fig. 2a). Meanwhile, P61 was mainly detected in the mitochondrial membrane (insoluble) fraction (Fig. 2a). As we have shown before, ORFH79 was also localized in the mitochondrial membrane (Peng et al., 2010). These data suggest that P61 and ORFH79 are possibly colocalized in vivo.
This work verified the interaction between ORFH79 and P61. To determine how ORFH79 causes CMS, we still needed to establish the function of P61 in mitochondria. Protein Blast showed that P61 had several homologs in plants (Fig. 2d). Two homologs from Arabidopsis and potato were reported to be a subunit of mitochondrion ETC complex III or cytochrome c reductase (Jansch et al., 1995; Meyer et al., 2008). In order to verify a similar location in rice, we used BN-PAGE to resolve the mitochondrion membrane-associated complex of YA and YB with detergent DDM (Fig. 2b). The bands of complex III were cut out and run through the second dimension Tricine-SDS-PAGE. Western blotting with antibody against P61 detected P61 signals in complex III of YA and YB (Fig. 2c). These results provided strong evidence that P61 was a subunit of ETC complex III. Meanwhile, ORFH79 signals were only detected in complex III of YA, as expected (Fig. 2c). These results showed P61 and ORFH79 colocalization in complex III of YA, which again matched well with previous evidence of their interaction.
Further, we observed the phenotype of the P61 mutant acquired from the Postech rice mutant library. The expression of P61 was disrupted by a T-DNA insertion in the 5′-UTR region. Except for the phenotype of abortive pollen, the mutant displayed significant defects in vegetative growth (see Fig. S2). This shows that the loss of function of P61 would seriously affect the normal growth of rice.
ORFH79 interacts with P61 in vivo
In the above experiments we showed that ORFH79 interacted with P61 by the bacterial two-hybrid system selection and also provided evidence that both ORFH79 and P61 were localized in complex III in the mitochondrial membrane fraction. In order to provide further evidence of a physical interaction between these two proteins in mitochondria of CMS line YA, bimolecular fluorescence complementation (BiFC) experiments in rice protoplasts were performed. Because ORFH79 is encoded by the mitochondrial genome in sterile line YA, it probably lacks the transit peptide that helps to transport it into mitochondria when it is transiently expressed in rice protoplasts. Thus, a transit peptide of ATPγ was added before the N terminus of ORFH79 to help its transport into mitochondria. As shown in Fig. 3(a), the positive yellow fluorescent protein (YFP) signals were detected only when the γH79-YC and P61-YN plasmids were cotransformed. Meanwhile, the YFP signals merged well with the mitochondrial RFP signals by plasmid FA-RFP (Fig. 3a). These results not only provided further evidence for their interaction, but also illustrated that the interaction happened inside the mitochondrion.
A coIP assay was performed to verify this interaction in vivo. Polyclonal antibody (αP61) against the C terminus of P61 was developed to characterize the interaction between P61 and ORFH79 (see the 'Materials and Methods' section). The mitochondria lysates prepared from YA and YB were first immunoprecipitated with antibody against P61. The input, immunoglobulin G (IgG) control, and precipitation were then immunoblotted with anti-ORFH79 antibody. As shown in Fig. 3(b), ORFH79 was detected in the precipitate of YA pulled down by anti-P61 antibody. As a control, ORFH79 was not detected in YB and IgG IP. On the other hand, when we used the anti-ORFH79 antibody to do the IP and the anti-P61 to do the western blot, P61 could also be detected in the precipitate of YA pulled down by anti-ORFH79 antibody (Fig. 3b). CoIP results showed that the precipitate pulled down both of the proteins using either of the two corresponding antibodies. In agreement with the bacterial two-hybrid system screening and colocalization data, both BiFC and coIP assays proved their interaction in vivo.
ORFH 79 inhibits activity of complex III
Subsequent comparison of P61 and its homologs, the QCR10 (6.5-kDa) subunit of complex III from beef and the QCR11 (8.5-kDa) subunit of complex III from yeast, showed that they exhibit significant sequence identity. Further analysis of the peptide sequence of P61 and its homologs (Table S2) on Interpro (http://www.ebi.ac.uk/) and PFAM (http://pfam.sanger.ac.uk/) revealed that they belong to a conserved family that currently has no name. The central part of this protein family consists of a transmembrane region (Fig. 4a). Based on this, we referenced the homologous structures from yeast (Brandt et al., 1994) and beef (Iwata et al., 1998) to predict the secondary structure of P61. The secondary structures of P61 and yeast QCR10 obey the ‘positive inside rule’ (Von Heijne & Gavel, 1988). As shown in Figs 4(a) and 2(d), a hydrophobic helix is in the middle of the proteins, which is flanked by an N-terminal domain with positively charged residues and an C-terminal domain with negatively and positively charged residues. We inferred that the N-terminal loop segment of P61, comprising amino acids 1–21, might be exposed to the matrix (inside), and the C-terminal loop segment, comprising amino acids 43–61, might be exposed to the intermembrane space. In order to determine which domain was involved in the interaction with ORFH79, two polypeptides (amino acids 1–24 and 41–61) were synthesized to perform the far western blot. The results showed that ORFH79 binds with the N-terminus loop segment of P61, but not with the C-terminus (Fig. 4b).
Since ORFH79 binds to a subunit of complex III, a possible band shift in BN-PAGE was estimated. However, we did not find significant band shift and quantity changes of complex III or the other complex between YA and YB in BN-PAGE gels (Fig. 2b). Histochemical staining of mitochondrial respiratory chain complexes on BN-PAGE gels showed that there was no significant difference in the enzyme activities of complex I, II, IV, and V (results not shown). Currently, there is no feasible method to stain complex III on BN-PAGE gel, so we tested the enzyme activity of complex III in solution. The results showed that there was a significant decrease of complex III activity in the mitochondria of CMS line YA when compared with that in the mitochondria of maintainer line YB (Fig. 5a). As a control, we tested the enzyme activity of complex V in solution. Consistent with the data from BN-PAGE gel staining, we observed no significant difference between YA and YB (Fig. 5b). We also designed an in vitro assay to test the impact of ORFH79 on normal mitochondria. After incubation with purified ORFH79 protein, normal mitochondria (YB) lysates also showed decreased complex III activity compared with the control (Fig. 5e). Therefore, we inferred that the binding of ORFH79 to P61 may specifically impair the enzyme activity of complex III, but does not affect its macromolecular composition.
In addition, we also compared the ATP and ROS content between purified YA and YB mitochondria. The ATP content in YA was significantly lower than in YB (Fig. 5c). By contrast, the ROS content in YA was 20% higher than in YB (Fig. 5d).
Theoretically, these physiological changes could happen when the mitochondrion became dysfunctional with regards to electron transport. Hence, we inferred that the binding between ORFH79 and complex III impaired the efficiency of electron transport and then slowed down ATP production. Moreover, increased electron leakage from ETC would produce more ROS as a result of ETC dysfunction.
P61 and orfH79 show increased expression in male reproductive tissue in CMS line
In order to detect the expression patterns of P61 and orfH79, we performed semiquantification PCR to test their expression in seven different tissues of YA and YB. P61 and orfH79 exhibited almost identical expression patterns in root, stem, young leaf, stigma, and anthers at the unicellular, bicellular, and matured stages in YA (Fig. 6a). Meanwhile, both genes showed a significant increase in expression in the unicellular- and bicellular-staged anthers. The coexpression pattern of P61 and OrfH79 again support their interaction in vivo. The expression pattern of P61 in YB was basically same as its pattern in YA (Fig. 6a). The expression pattern of P61 is also illustrated in Fig. 6(b) based on microarray data (Fujita et al., 2010). The results showed that P61 was also constitutively expressed in all tested tissues and exhibited a significant increase (more than two times) in anthers at the bicellular and tricellular stages. We randomly chose other ETC subunits to study their expression based on the same microarray data. These genes, except for one subunit of complex V, exhibited nearly the same expression pattern as P61 (Fig. 6b).
Expression of cell cycle-related genes in pollen was affected in CMS line
The phenotype of CMS presents as pollen abortion, which is caused by the constraint on the division of pollen generative cells. Honglian CMS aborted at the binucleate stage, meaning that the expression of cell cycle-related genes in pollen may be reprogrammed under CMS background. Recent studies in Arabidopsis have made possible the construction of a cell cycle regulatory pathway during the secondary pollen mitosis (Fig. 7c; Iwakawa et al., 2006; Nowack et al., 2006; Harashima et al., 2007; Kim et al., 2008; Dissmeyer et al., 2009; Gusti et al., 2009), and many nuclear genes are involved in this process (Twell et al., 2002, 2005; McCormick, 2004; Twell, 2006, 2011; Zhang & Wilson, 2009). We searched for the homologs of these genes in rice and compared their expression at different microspore developmental stages between sterile line YA and maintainer line YB. Real-time PCR results showed that three of the CDKA;1 homologs (Loc_Os02 g03060, Loc_Os03 g01850, and Loc_Os03 g02680) in rice showed a significant decrease in sterile line YA from the unicellular stage (Fig. 7a). Western blot results also illustrated that the amount of total CDKA1 and phosphorylated CDKA1 decreased in YA, which was more obvious at the bicellular stage (Fig. 7b). Furthermore, the KRP6/7 homolog (Loc_Os11 g40030), which was the phosphorylation inhibitor of CDKA1, increased to varying degrees at the RNA and protein levels, especially at the bicellular stage (Fig. 7a,b). Meanwhile, as putative E3 ligase components that are involved in degrading KRP6/7 through the ubiquitin-proteasome pathway, RHF homolog (Loc_Os03 g57500) and FBL17a homolog (Loc_Os03 g43390) exhibited a decrease in YA from the unicellular stage (Fig. 7a). These data imply that the CMS protein ORFH79 ultimately causes abortive pollen through the ubiquitin-proteasome pathway that regulates cell cycle-associated genes in pollen.
The plant trait of CMS is usually determined by a mitochondrial gene and is recognized by the sterile pollen phenotype. Understanding the underlying mechanism of CMS is the basis for comprehending natural selection forces of CMS (Budar et al., 2003). When summarizing current studies on plant CMS, it is hard to unify our view on the mechanism of CMS. Both ‘gain-of-function’ and ‘loss-of-function’ models are supported by a lot of experimental evidence. Our study on Honglian CMS rice supports the loss-of-function model.
In our experiment, the macromolecular composition of these mitochodrial complexes resolved by BN-PAGE showed almost no difference between sterile lines and maintainer lines (Fig. 2b). Meanwhile, the ETC complex III activity of CMS lines was c. 45% lower than that of maintainer lines (Fig. 5a). These seemingly contradictory results in fact imply that ORFH79 binding to P61 reduced enzyme activity of ETC complex III, without affecting the overall composition of this complex. How this binding reduces enzyme activity is still unknown. One possibility is that the binding results in the malfunction of P61, which then leads to the decrease in enzyme activity. In yeast, the deletion of QCR10 (P61 homolog gene) significantly reduces the activity of ETC complex III; however, its deletion also significantly affects the composition of ETC complex III (Brandt et al., 1994). However, the HL-CMS lines and its maintainer basically showed no significant difference in vegetative growth. Thus, the deletion or RNAi of P61 in rice did not simulate the real situation of HL-CMS lines. We infer that ORFH79 somehow binds to P61 to affect the electron transfer in an unknown way, but does not affect the structure of the complex III dimer. Recent studies of plant mitochondrial supercomplexes indicate that the respiratory chain complexes (complexs I–IV) and the ATP synthase complex (complex V) are organized into different supercomplexes (Eubel et al., 2003; Dudkina et al., 2006; Rasmusson et al., 2008). ORFH79 may affect the formation of a supercomplex in which complex III participates.
Cytoplasmic male sterility plants are characterized by their inability to produce functional pollen. Interestingly, the CMS genes only disrupt pollen development, but vegetative development and female fertility are apparently unaffected. Previous work in our laboratory demonstrated that ORFH79 in HL-CMS could be detected in mitochondria from callus, root, leaf, and spikelet of CMS lines (Peng et al., 2010). The present results showed that P61 and OrfH79 had a coexpression pattern in seven different tissues of YA and enhanced expression at the post-meiosis stage of anther development (Fig. 6a). The analysis of microarray data indicates that the mitochondrial ETC genes accumulate during the mitosis stage of pollen (Fig. 6b). These results might explain why orfH79 only causes abnormal development of microspores. The leading hypothesis for CMS is that the developing microspores consume much more energy than any other tissues. The number of mitochondria increases significantly in the microspores and tapetum of the anther after the small gametophytic meiosis stage (Lee & Warmke, 1979). Some abnormal phenotypes caused by impairment of mitochondrial function also occur at this stage (Hanson & Bentolila, 2004). Two recent studies reporting that a mutation in flavoprotein subunit of complex II (Leon et al., 2007) and a mutation in a subunit of complex V (Li et al., 2010) both specifically cause abortive microspores support this hypothesis. Our findings here provide further supporting evidence. In fact, the microspores of HL-CMS rice could not enter into the second stage of mitosis, as they stopped at the bicellular stage (Fig. S1; Li et al., 2007). Although the microspores of HL-CMS could pass the meiosis and first mitosis stage, they showed abnormal mitosis during the highest energy-consuming stage (P2 and P3 in Fig. 6b) by the impairment of ORFH79.
Mitochondrial signals can influence the expression of nuclear genes, which is called mitochondrial retrograde regulation (Woodson & Chory, 2008). Although the energy limitation hypothesis more likely explains the tissue-specific injury caused by ORFH79 in HL-CMS, it is still not clear how the status of energy limitation leads to suspension of mitosis of pollen. In other words, the retrograde regulation of mitochondria in HL-CMS still requires resolution. ETC is considered an important source of oxygen radicals in mitochondria (Moller, 2001; Petrosillo et al., 2003; Brand, 2010). The impairment of complex III activity, as a result of ORFH79 binding, may increase the electron leakage from the electron transport chain to generate excessive ROS while reducing ATP production. Thus the excessive ROS and reduced ATP are likely the primary signals to be transferred to the nucleus, and ultimately lead to abnormal mitosis during the development of microspores (Moller & Sweetlove, 2010). Coincidently, a study in Drosophila melanogaster showed that shortage of ATP and overdose of ROS caused by disruption of the mitochondrial electron transport chain acted as mitochondrial retrograde signals to activate the G1-S checkpoint (Owusu-Ansah et al., 2008; Mandal et al., 2010). In the present paper, we investigated the expression of cell cycle-related genes in YA and YB. Their changes in YA imply that the pollen of HL-CMS may activate the G1-S checkpoint, as CDKA;1, the key regulator of the G1-S checkpoint of mitosis, exhibited a decrease in expression at RNA, protein, and protein modification (phosphorylation) levels (Fig. 7b).
Based on these reports and our findings, we propose a hypothesis for the molecular mechanism of HL-CMS, as summarized in Fig. 8. CMS protein ORFH79 binds to ETC complex III and ultimately causes increased ROS and reduced ATP concentrations. This status of mitochondria is then sensed by the nucleus through a retrograde pathway. This information transferred from the organelle (mitochondrion) forces the nucleus to stop at the G1-S checkpoint of the cell cycle. As a result, the pollens of Honglian CMS rice exhibit an abortive phenotype.
Combined with the significantly different enzyme activity of ETC complex III between the CMS line and the maintainer line, we infer that ORFH79 binds to ETC complex III via P61, which subsequently suppresses its activity and ultimately causes the pollen abortion phenotype.
We thank Hwang, Inhwan (Pohang University of Science and Technology, Korea) for kindly providing the FA-RFP vector. This work was partly supported by the National 973 Project (2007CB109005, 2011CB100102) and the National Natural Science Foundation (31070303) of China.