A Nuclear restorer-of-fertility Mutation Disrupts Accumulation of Mitochondrial ATP Synthase Subunit in Developing Pollen of S Male-Sterile Maize

Corresponding author: Christine D. Chase, 1301 Fifield Hall, University of Florida, Gainesville, FL 32611., ctdc{at}mail.ifas.ufl.edu (E-mail)

Communicating editor: J. BIRCHLER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Mitochondrial biogenesis and function depend upon the interaction of mitochondrial and nuclear genomes. Forward genetic analysis of mitochondrial function presents a challenge in organisms that are obligated to respire. In the S-cytoplasmic male sterility (CMS-S) system of maize, expression of mitochondrial open reading frames (orf355-orf77) conditions collapse of developing haploid pollen. Nuclear restorer-of-fertility mutations that circumvent pollen collapse are often homozygous lethal. These spontaneous mutations potentially result from disruption of nuclear genes required for mitochondrial gene expression, in contrast to homozygous-viable restorer-of-fertility alleles that function to block or compensate for the expression of mitochondrial CMS genes. Consistent with this hypothesis, the homozygous-lethal restoring allele historically designated RfIII was shown to be recessive in diploid pollen produced by tetraploid CMS-S plants. Accordingly, the symbol for this allele has been changed to restorer-of-fertility lethal 1 (rfl1). In haploid rfl1 pollen, orf355-orf77 transcripts and mitochondrial transcripts encoding the -subunit of the ATP synthase (ATPA) were decreased in abundance. Haploid rfl1 pollen failed to accumulate wild-type levels of ATPA protein, indicating that functional requirements for mitochondrial protein accumulation are relaxed in maize pollen. The CMS-S system and rfl mutations therefore allow for the selection of nuclear mutations disrupting mitochondrial biogenesis in a multicellular eukaryote.

MITOCHONDRIAL genomes encode a small subset of the proteins required for mitochondrial function. Organelle targeting prediction programs and mitochondrial proteomics studies indicate that 10% of eukaryotic nuclear genes encode products targeted to mitochondria (GRIVELL et al. 1999 ; EMANUELSSON et al. 2000 ; KARLBERG et al. 2000 ; KRUFT et al. 2001 ). Because mutations disrupting mitochondrial respiration confer lethality or severe developmental abnormalities in obligate aerobes (NEWTON and GABAY-LAUGHNAN 1998 ; SHOUBRIDGE 2001 ; WALLACE 2001 ), systematic forward genetic analysis of mitochondrial function has been accomplished primarily in the facultative anaerobe Saccharomyces cerevisiae (GRIVELL et al. 1999 ). While a powerful model (STEINMETZ et al. 2002 ), the yeast mitochondrial proteome lacks the complexity observed in, and conserved between, plants and mammals (EMANUELSSON et al. 2000 ; KARLBERG et al. 2000 ; KRUFT et al. 2001 ).

Plant cytoplasmic male sterility (CMS) systems provide tools to investigate mitochondrial-nuclear genome interactions in multicellular eukaryotes (HANSON 1991 ; SCHNABLE and WISE 1998 ; WISE and PRING 2002 ). Expression of mitochondrial CMS genes conditions failure to produce functional pollen. Male fertility results from interactions between nuclear restorer-of-fertility alleles and mitochondrial CMS genes. Two restorer-of-fertility genes have been cloned and functionally characterized. The Rf2 allele is required, along with the unlinked Rf1 allele, to condition male fertility in T-cytoplasm maize (DUVICK 1965 ). Analysis of the cloned rf2 locus indicates that the functional Rf2 allele encodes a mitochondrial aldehyde dehydrogenase, which in some way compensates for the expression of the mitochondrial URF13 protein that causes male sterility (CUI et al. 1996 ; LIU et al. 2001 ). In Petunia CMS, a restoring allele designated Rf1 encodes a mitochondrial-targeted, pentatricopeptide repeat protein that disrupts accumulation of the mitochondrial protein responsible for male sterility (NIVISON and HANSON 1989 ; BENTOLILA et al. 2002 ). In both CMS petunia and CMS-T maize, male fertility is determined by mitochondrial-nuclear interactions that occur in diploid (2n) anther tissues (WARMKE and LEE 1977 ; CONLEY and HANSON 1995 ). Because the Petunia Rf1 allele and maize Rf1 and Rf2 alleles are all dominant, 2n tissues heterozygous for restoring and nonrestoring alleles are fully rescued.

In CMS-S maize, male sterility and fertility are determined by mitochondrial-nuclear interactions in the haploid male gametophyte (BUCHERT 1961 ). Cotranscribed mitochondrial reading frames (orf355 and orf77) have been implicated in S male sterility. Although protein expression is yet to be confirmed, a 1.6-kb orf355-orf77 transcript is always absent after cytoplasmic reversion of S male sterility to male fertility (ZABALA et al. 1997 ). CMS-S pollen carrying a restoring allele develops normally, whereas pollen carrying a nonrestoring allele collapses after the microspore mitosis (BUCHERT 1961 ; LEE et al. 1980 ). Thus, in S cytoplasm, only restoring alleles transmit through the pollen.

The Rf3 restoring allele, which occurs naturally in the maize inbred lines CE1 and Ky21 (DUVICK 1965 ), cosegregates with processing within orf355-orf77 transcripts in developing CMS-S pollen (WEN and CHASE 1999A ). This observation is consistent with a fertility restoration mechanism observed in multiple CMS systems: the processing of CMS locus transcripts and concomitant reduction in the accumulation of CMS-specific proteins. Alleles at these loci often cosegregate with processing events in normal mitochondrial gene transcripts (reviewed by WISE and PRING 2002 ). Restoring alleles of this type therefore reflect genetic variation at nuclear loci that function in normal mitochondrial RNA processing pathways.

Because fertility restoration for CMS-S maize occurs in haploid tissues, dominance relationships of restoring and nonrestoring alleles are not necessarily known. A restoring allele might function to block or compensate for the expression of the mitochondrial CMS locus. Alternatively, loss of a nuclear function essential to expression of CMS could also condition male fertility. In the case of the rf3 locus, dominance of the restoring (Rf3) allele was demonstrated by cotransmission of restoring and nonrestoring alleles through 2n pollen produced by CMS-S tetraploid (4n) plants (KAMPS et al. 1996 ). This may not be the case for all restoring alleles of S cytoplasm. Spontaneous mutations that circumvent CMS-S pollen collapse are recovered at multiple nuclear loci. These mutations were historically given an Rf designation on the basis of the male-fertile phenotype conferred by the mutation. Many restorer-of-fertility mutations, however, are homozygous lethal with respect to seed development (LAUGHNAN and GABAY 1975 , LAUGHNAN and GABAY 1978 ; GABAY-LAUGHNAN et al. 1995 ). Restorer-of-fertility mutations may therefore represent recessive, loss-of-function alleles in nuclear genes required for the expression of essential mitochondrial genes. Such mutations, however, would be readily observed in haploid pollen.

The loss-of-function model for restorer-of-fertility mutations makes two testable predictions. First, restoring alleles recovered by spontaneous mutation should be recessive to nonrestoring alleles with respect to fertility restoration. Second, restorer-of-fertility mutations may disrupt accumulation of mitochondrial-encoded gene products essential to respiratory function. To test the proposed model, we studied the homozygous-lethal restoring allele historically designated RfIII (LAUGHNAN and GABAY 1973 , LAUGHNAN and GABAY 1975 ). Dominance relationships of restoring and nonrestoring alleles at this locus were examined in 2n pollen produced by 4n CMS-S plants. The effects of the restoring allele on mitochondrial gene expression were examined in developing haploid pollen. We present evidence for the recessive nature of the restoring allele in 2n pollen and evidence that the restoring allele conditions failure to accumulate wild-type levels of mitochondrial ATP synthase subunit (ATPA) protein in haploid pollen. Accordingly, this restorer-of-fertility allele has been designated restorer-of-fertility lethal 1 (rfl1). Because homozygous-lethal, restorer-of-fertility mutations also occur at other maize nuclear loci (LAUGHNAN and GABAY 1978 ; GABAY-LAUGHNAN et al. 1995 ), rfl1 becomes the founding member in a series of rfl loci. The viability of rfl1 pollen indicates that maize pollen has relaxed requirements for mitochondrial protein accumulation and that rfl mutations provide a means to select for mutations disrupting mitochondrial biogenesis in a multicellular eukaryote.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Maize genetic nomenclature:
In maize genetic nomenclature (http://www.agron.missouri.edu/maize_nomenclature.html), loci are indicated by lowercase italics (e.g., the rf3 locus). Alleles at a locus are also indicated in italics, with the first letter capitalized for dominant alleles (e.g., the Rf3 allele). Dominance relationships of restoring and nonrestoring alleles that act in haploid pollen cannot be determined by direct observation of pollen phenotype (KAMPS et al. 1996 ). For restorer-of-fertility alleles that are homozygous lethal, we abandon the conventional Rf and rf designations for restoring and nonrestoring alleles, respectively. On the basis of the work presented below, the conventional designations do not reflect correct dominance relationships. We adopt the designation restorer-of-fertility lethal (rfl) for homozygous-lethal restoring alleles. The symbol rfl1 replaces the previously published symbol RfIII (LAUGHNAN and GABAY 1973 , LAUGHNAN and GABAY 1975 ). Because multiple, independent rfl mutations have been recovered and mapped to different sites (LAUGHNAN and GABAY 1978 ; GABAY-LAUGHNAN et al. 1995 ), a series of rfl loci is anticipated.

Plant materials and genetic manipulations:
The plant materials used in this work are summarized in Table 1. The homozygous-lethal rfl1 allele arose by spontaneous mutation in the R853 inbred background. The mutation was recovered from a male-fertile plant in a population of male-sterile CMS-S plants (LAUGHNAN and GABAY 1973 ). The source of the homozygous-viable Rf3 allele was inbred line CE1. The rfl1 and Rf3 alleles were each moved into the S-cytoplasm version of inbred Mo17 (Mo17-S) by a minimum of six backcross generations prior to use in this study. All R75 4n plants used in this study descended from self-pollination of a single, N-cytoplasm, R75 4n plant. To determine the status of these R75 4n materials with respect to S-cytoplasm restorer-of-fertility alleles, a CMS-S W23 4n without restoring alleles developed by KAMPS et al. 1996 was pollinated with N-cytoplasm R75 4n. If this N-cytoplasm R75 4n plant carried a single, dominant restoring allele for S cytoplasm, one-quarter of the progeny from this cross would be expected to inherit this allele and to exsert anthers and shed pollen. None of the 26 progeny examined exhibited anther exsertion or pollen shed. Therefore, the R75 4n parent most probably did not carry a dominant restoring allele for S cytoplasm. Indeed, no evidence of any dominant restoring allele was found in materials subsequently developed from this parent. The presence of recessive restoring alleles in the R75 4n parent cannot be ruled out, as these would not condition male fertility in 2n pollen of our testcross progeny. Recessive restoring alleles can arise by spontaneous mutation at many different loci (LAUGHNAN and GABAY 1978 ; GABAY-LAUGHNAN et al. 1995 ), so any recessive restoring allele that might be carried by R75 4n is not likely to be at the rfl1 locus or to confound our analysis of restoring and nonrestoring alleles at that locus.

CMS-S 4n plants carrying restoring and nonrestoring alleles at the rfl1 locus were constructed as described by KAMPS et al. 1996 . Briefly, 2n plants homozygous for the elongate (el) mutation were pollinated by a 4n plant. The el/el genotype results in occasional unreduced female gametes, which can be rescued as viable seed after fertilization by 2n pollen (RHOADES and DEMPSEY 1966 ). The entire crossing scheme and predicted progeny are outlined in Fig 1A.

View larger version (37K): In this window In a new window Download PPT slide   Figure 1. Construction and RFLP analysis of CMS-S 4n plants. (A) The crosses performed to construct CMS-S 4n plants are outlined. For simplicity, only the genotypes carried forward from each generation are shown. The crossing parents are described in Table 1. (B) RFLP analysis at the whp1 locus of CMS-S 4n plants and their progenitors. Total cellular DNAs were digested with SstI prior to electrophoresis, blotting, and hybridization with the whp1 probe. Each lane contains 50 µg of total nucleic acids. DNA sources are indicated above the lanes. Lanes 4 and 5 contain DNA from the 2n female and 4n male parents, respectively, of the 4n progeny plants analyzed in lanes 6–16. SstI digestion distinguished whp1 alleles from each of the progenitor lines (W192BN CA, 805E, the rfl1 source, and R75 4n) as indicated to the right of the image captured from the autoradiograph.

Male fertility analysis:
Florets were removed from the middle third of the main rachis on fully emerged tassels prior to anther exsertion. Florets were stored in 70% ethanol. Microspore mitosis is complete in florets of a fully emerged tassel (CHANG and NEUFFER 1989 ). S-cytoplasm pollen without restoring alleles collapses abruptly after microspore mitosis (LEE et al. 1980 ; WEN and CHASE 1999B ) and has collapsed by the time a tassel has fully emerged (our unpublished observations). Anthers were dissected from the florets and squashed in acetocarmine stain (SISCO et al. 1985 ) on a microscope slide. Pollen grains were visualized under a light microscope at x100 magnification; 355–600 grains were counted for each plant. Rounded grains were scored as starch filling; shrunken grains were scored as collapsed. The percentage of starch-filling grains was calculated for each sample.

To test for an effect of genotype at the rfl1 locus on the 2n pollen phenotype, the genotype of individual CMS-S 4n plants at rfl1 was determined by analyzing the restriction fragment length polymorphism (RFLP) genotype at the linked whp1 locus as described below. Because the same number of pollen grains was not counted for each plant, the number of starch-filling grains in each pollen sample was normalized to a sample of 355, the minimum number counted for any plant. Within a segregating family, the mean number of starch-filling pollen grains in a sample of 355 was calculated for 4n plants having two restoring alleles (Rfl1/Rfl1/rfl1/rfl1), one restoring allele (Rfl1/Rfl1/Rfl1/rfl1), or no restoring alleles (Rfl1/Rfl1/Rfl1/Rfl1). The means were then compared by a Student's t-test (STUDENT 1908 ) that employed a correction in the degrees of freedom to adjust for the unequal sample sizes and variances (HAYES 1973 ).

DNA extractions and Southern blots:
DNAs were extracted and analyzed for RFLPs as described by KAMPS et al. 1996 . The whp1 probe was obtained from U. Weinand and other probes were obtained from the Missouri Maize Project (http://www.agron.missouri.edu/probes.html).

Pollen RNA and protein analysis:
The recovery of microspores and pollen at different stages of development was accomplished by sucrose density gradient centrifugation as described by WEN and CHASE 1999B . The extraction, electrophoresis, transfer, and hybridization of total cellular RNA samples were also performed as described by WEN and CHASE 1999B . SDS-soluble proteins were extracted by mixing 0.25-g samples of starch-filling pollen (separated from collapsed pollen by sucrose density gradient centrifugation) with 300 µl of SDS sample buffer. Following a 10-min incubation at 95°, insoluble material was pelleted by centrifugation for 10 min at 16,000 x g in a microcentrifuge. Supernatants were removed and stored at -20°. Supernatant protein concentrations were determined with the CoomassiePlus protein assay reagent (Pierce Endogen, Rockford, IL). Proteins were fractionated by SDS polyacrylamide gel electrophoresis, transferred to nitrocellulose, and labeled with monoclonal antibodies recognizing COXII or ATPA or with polyclonal antibodies recognizing NAD9. Antigen-antibody complexes were labeled with horseradish peroxidase (HRP)-conjugated anti-mouse or anti-rabbit IgG. Complexes were visualized following reaction with chemiluminescent HRP substrate (Pierce Endogen) and exposure of the blot to X-ray film. Antibodies were removed with Restore Western stripping buffer (Pierce Endogen) prior to labeling with another primary antibody.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Identification of RFLP markers linked to the rfl1 locus:
To facilitate subsequent genetic studies, we identified molecular markers linked to the rfl1 locus. Previous crosses with wx1-linked translocations (LAUGHNAN and GABAY 1978 ) placed the rfl1 locus on chromosome 2 (GABAY-LAUGHNAN et al. 1995 ). Chromosome 2 RFLP markers linked to rfl1 were identified through analysis of a backcross (BC) population (W182BN-CA//W182BN-CA/Mo17 S rfl1/Rfl1, where CA is a subgroup of the S cytoplasm) by the strategy of KAMPS and CHASE 1997 . All 34 BC progeny carried the rfl1 allele, indicated by a normal tassel phenotype, because nonrestoring (Rfl1) alleles do not generally transmit through pollen of CMS-S plants. Microscopic examination of pollen samples showed the expected phenotype of 50% starch-filling pollen. RFLP loci unlinked to the rfl1 locus were identified by independent (1:1) segregation for W182BN/W182BN and W182BN/Mo17 genotypes with randomly selected individuals in the population. RFLP loci were determined to be linked to the rfl1 locus if the population contained predominantly or exclusively the W182BN/Mo17 genotype. Segregation data for seven chromosome 2 RFLP loci are summarized in Table 2. Of these, the whp1 locus on the long arm of chromosome 2 exhibited the tightest linkage to the rfl1 locus with no recombinants observed among the full set of 34 BC individuals.

Construction and fertility analysis of CMS-S 4n plants:
CMS-S 4n plants carrying restoring and nonrestoring alleles at the rfl1 locus were constructed to examine the dominance relationships of these alleles in 2n pollen. The homozygous lethality of the rfl1 allele suggests that this allele should also be recessive with respect to fertility restoration. Six families of 4n plants were generated as shown in Fig 1A, two with the R75 4n male parent and four with the W23 4n male parent. The resulting CMS-S 4n families were predicted to include three different genotypes at the rfl1 locus. All progeny must carry two nonrestoring (Rfl1) alleles inherited from the R75 4n or W23 4n parent in addition to various combinations of Rfl1 and rfl1 alleles from the 2n parent. Because the el/el genotype conditions unreduced female gametes at variable frequency (RHOADES and DEMPSEY 1966 ), the frequency of each 4n genotype is expected to vary among families.

In the case of the families derived by fertilization of unreduced female gametes with R75 2n pollen, the recovery of the expected 4n genotypes could be verified by RFLP analysis with the whp1 probe (Fig 1B). The whp1 alleles linked to restoring (rfl1) and nonrestoring (Rfl1) alleles of the 2n parents and grandparents could be distinguished from each other and from the whp1 allele of the R75 4n parent. Genotyping was not possible in the case of the W23 4n-derived families, as the whp1-W23 4n allele (not shown) could not be distinguished from the whp1-rfl1 source allele.

Fertility analysis of 4n plants can be used to distinguish dominance relationships of restoring and nonrestoring alleles at the rfl1 locus. The predicted genotypes and genotype frequencies of 2n pollen produced by the 4n plants are summarized in Table 3. CMS-S pollen homozygous for nonrestoring alleles (Rfl1/Rfl1) is predicted to collapse, and CMS-S pollen homozygous for restoring alleles (rfl1/rfl1) is predicted to develop normally. Dominance relationships are therefore determined by the phenotype of CMS-S pollen with the heterozygous genotype (Rfl1/rfl). The phenotype of CMS-S Rfl1/rfl1 2n pollen is predicted to be collapsed if the restoring (rfl1) allele is recessive with respect to fertility restoration. If the restoring allele at the rfl1 locus is dominant, the phenotype of CMS-S Rfl1/rfl1 pollen is predicted to be starch filling. In the case of a recessive restoring allele, only the homozygous rfl1/rfl1 pollen will function. The predicted frequency of rfl1/rfl1 pollen (4–21%) might be insufficient to result in pollen shed. Thus, a recessive restoring allele predicts a male-sterile phenotype for the 4n plants carrying one or two restoring (rfl1) alleles. In the case of a dominant restoring allele, all plants carrying restoring alleles are expected to shed pollen (Table 3), as observed for the dominant Rf3 allele (KAMPS et al. 1996 ).

The R75 4n-derived families were analyzed for male fertility and for genotype at the whp1 locus. Each family was grown and evaluated for fertility under both fall and spring Florida field conditions. None of the plants shed pollen, although most exserted scattered, thin anthers. This phenotype is common among S male-sterile materials (DUVICK 1965 ). Pollen counts for each plant were determined from anther squashes, and each plant was genotyped for RFLPs at the rfl1-linked whp1 locus. The pollen phenotypes observed for each plant genotype are summarized in Table 4.

Fertility analysis of these two 4n families (Table 4) was consistent with a recessive restoring allele at the rfl1 locus. Anthers from plants homozygous for the nonrestoring allele, genotype Rfl1/Rfl1/Rfl1/Rfl1 as determined by genotype at the linked whp1 locus, contained higher-than-expected frequencies of starch-filling pollen, given that anthers of control CMS-S W23 4n plants and CMS-S W23 4n/R75 4n plants grown in the same environment contained 1–6% and 1–10% starch-filling pollen, respectively. Pollen counts in the informative genotypes were therefore confounded by a high background of pollen that failed to collapse for unknown reasons. Indeed, counts of starch-filling pollen for some plants having one restoring allele (genotype Rfl1/Rfl1/Rfl1/rfl1) exceeded the predictions for a recessive restoring allele. However, only four of these plants (all in family 45-18) contained frequencies of starch-filling pollen consistent with a dominant restoring allele, and none of those plants shed pollen. Furthermore, anthers from the five family 45-18 plants having two restoring alleles (genotype Rfl1/Rfl1/rfl1/rfl1) contained 11–34% starch-filling pollen. These values were a close match to the 10–32% starch-filling pollen observed in anthers of siblings having no restoring alleles (genotype Rfl1/Rfl1/Rfl1/Rfl1) and did not approach the 78% starch-filling pollen predicted for the case of a dominant restoring allele at the rfl1 locus (Table 3). Finally, a Student's t-test supported the conclusion that the mean number of starch-filling pollen grains in samples collected from the plants without restoring alleles did not differ (at the 0.05 level of significance) from the mean number of starch-filling pollen grains in samples collected from plants having one or two restoring alleles.

While members of the W23 4n-derived families could not be directly genotyped with respect to the whp1 locus, W23 4n-derived plants exhibited phenotypes similar to plants in the genotyped 4n families. Twenty plants from each of four W23 4n families were examined for tassel and pollen phenotypes. These plants exserted scattered thin anthers that shed no pollen. The mean percentages (and ranges) of starch-filling pollen observed in these families, 13% (1-26), 9% (1-44), 20% (8-41), and 29% (4-52), were similar to those observed in the R75 4n-derived families.

Effects of restorer-of-fertility alleles on mitochondrial transcript accumulation:
The rfl1/rfl1 genotype is lethal for seed development. The rfl1 allele is therefore predicted to influence the expression of both CMS and essential mitochondrial gene loci. The effects of rfl1 and Rf3 restoring alleles on mitochondrial gene expression were examined in developing haploid pollen. Mitochondrial gene expression was first examined at the RNA level for the CMS-associated orf355-orf77 locus. As S-cytoplasm Rf3 pollen progresses through the microspore and starch-filling stages of pollen development, the 1.6-kb orf355-orf77 transcript exhibits decreased accumulation and a 1.1-kb orf355-orf77-hybridizing transcript appears (WEN and CHASE 1999A ; Fig 2). Decreased accumulation of the 1.6-kb orf355-orf77 transcript also occurred in S-cytoplasm rfl1 pollen development, but without the coinciding appearance of shorter transcripts (Fig 2). To determine whether the rfl1 locus has a role in the expression of other mitochondrial genes, we hybridized RNA blots with the mitochondrial 18S rRNA (rrn18) probe and with probes for mitochondrial genes encoding respiratory complex subunits (atpa, atp6, atp9, cob, coxI, and nad9). Relative to rrn18 transcripts, only atpa transcripts exhibited decreased accumulation in S-cytoplasm rfl1 pollen as compared to S-cytoplasm Rf3 pollen (Fig 2).

View larger version (46K): In this window In a new window Download PPT slide   Figure 2. Mitochondrial transcript accumulation in S-cytoplasm pollen development. RNA blot hybridizations were performed following fractionation of total cellular RNA extracted from microspores (MSP) at early (1, 7), mid (2, 8), and late (3, 9) stages of development and from starch-filling pollen (SFP) at early (4, 10), mid (5, 11), and late (6, 12) stages. Microspores and starch-filling pollen were recovered from both Mo17-S Rf3/rf3 (lanes 1–6) and Mo17-S Rfl1/rfl1 (lanes 7–12) plants. Microspore preparations contained a mixture of restoring and nonrestoring genotypes. Starch-filling pollen, physically separated from collapsed pollen by centrifugation on sucrose density gradients, included only the restoring genotype. All lanes contained 10 µg of RNA.

Effects of restorer-of-fertility alleles on mitochondrial protein accumulation:
Immunodecoration of SDS-soluble pollen proteins fractionated by polyacrylamide gel electrophoresis was used to investigate the accumulation of mitochondrial proteins in the presence of the rfl1 allele. This approach demonstrated failure of S-cytoplasm rfl1 pollen to accumulate wild-type levels of ATPA protein, in contrast to S-cytoplasm Rf3 pollen or N-cytoplasm pollen with no restoring alleles (Fig 3A). Another mitochondrial gene product, COXII, accumulated to similar levels in all pollen genotypes. Failure of the rfl1 pollen to accumulate ATPA was likely an effect of the rfl1 allele since all pollen genotypes carried the inbred Mo17 nuclear background and the same S cytoplasm.

View larger version (51K): In this window In a new window Download PPT slide   Figure 3. Mitochondrial protein accumulation during pollen development. (A) Decreased abundance of ATPA protein in rfl1 pollen. SDS-soluble proteins were extracted from starch-filling pollen, fractionated, and decorated with monoclonal antibodies recognizing COXII or ATPA. Pollen samples were from a normal (N) cytoplasm Mo17 plant without restoring alleles (1), a Mo17-S Rf3/rf3 plant (2 and 3), and five different Mo17-S Rfl1/rfl1 plants (lanes 4–8). (B) Cosegregation of rfl1 with reduced accumulation of ATPA. Starch-filling pollen (SFP) (samples 5–8, 13–16, and 21–24) was isolated from individual Rfl1/rfl1 plants and collapsed pollen (CP; samples 1–4, 9–12, and 17–20) was isolated from individual Rfl1/Rfl1 plants. SDS-soluble proteins were recovered and analyzed as described for A. Protein samples (30–40 µg of protein from collapsed pollen or 6–10 µg of protein from starch-filling pollen) were loaded to achieve similar COXII signals; COXII signals were subsequently removed and the blots labeled with ATPA antibodies. Similar results (not shown) were obtained when the order of antibody probing was reversed and when samples were loaded on the basis of signals obtained with antibodies recognizing the mitochondrial-encoded NAD9 protein.

Cosegregation analysis confirmed the linkage of rfl1 with the failure to accumulate ATPA protein. A Mo17-S Rfl1/rfl1 plant was pollinated with a Mo17-N Rfl1/Rfl1 plant, creating a population segregating for male-fertile (Mo17-S Rfl1/rfl1) and male-sterile (Mo17-S Rfl1/Rfl1) progeny. Pollen phenotypes of 38 plants were examined by light microscopy. The 18 male-sterile and 20 male-fertile plants confirmed the expected 1:1 segregation (Yates' corrected ² = 0.026, P > 0.90; YATES 1934 ). Tassels of 24 plants were processed individually to recover collapsed pollen from Rfl1/Rfl1 plants and starch-filling pollen (physically separated from collapsed pollen) of Rfl1/rfl1 plants. Examination of the starch-filling pollen preparations by light microscopy revealed <5% contaminating collapsed pollen in the starch-filling pollen samples. Fractionation and immunodecoration of SDS-soluble pollen proteins revealed perfect cosegregation of the rfl1 allele with failure to accumulate ATPA (Fig 3B).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic basis of the rfl1 mutation:
The linkage of the rfl1 and whp1 loci (Table 2) places the rfl1 locus in the region of the rf3 locus on the long arm of maize chromosome 2 (KAMPS and CHASE 1997 ) and raises the question of whether rfl1 and Rf3 are alleles at the same locus. Direct tests of allelism have established that rfl1 and rf3 are distinct genetic loci (S. GABAY-LAUGHNAN, unpublished observations). While both the rfl1 and Rf3 alleles reduce accumulation of mitochondrial orf355-orf77 transcripts (WEN and CHASE 1999A ; Fig 2), no evidence of RNA processing was observed in the case of rfl1. These transcript differences may reflect different mechanisms of fertility restoration.

The rfl1 allele is recessive with respect to embryo development. Embryos with the rfl1/rfl1 genotype fail to develop (LAUGHNAN and GABAY 1973 ) whereas Rfl1/rfl1 embryos inheriting the rfl1 allele from either parent develop normally. This observation supports the hypothesis that the rfl1 allele resulted from a loss-of-function mutation. The loss-of-function model predicts that the rfl1 restoring allele should be recessive to the nonrestoring allele in CMS-S 2n pollen. This model is supported by the observation that CMS-S 4n plants having one or two restoring (rfl1) alleles exhibit a male-sterile phenotype (Table 4). We believe that the male-sterile phenotypes of these plants are the result of genetic interactions between Rfl1 and rfl1 alleles and not the result of environmental effects or the failure of a dominant restoring allele to function in a given genetic background. Similar phenotypes were observed during fall and spring growing seasons in Florida and in both the R75 4n and W23 4n genetic backgrounds. The male sterility of CMS-S Rfl1/Rfl1/rfl1/rfl1 plants contrasts with the fertility analysis of CMS-S Rf3/Rf3/rf3/rf3 plants developed in the W23 4n background. The CMS-S 4n plants carrying two restoring (Rf3) alleles produced anthers that shed 78–92% starch-filled pollen, which in turn produced viable seeds upon fertilization of 4n females. RFLP analysis of resulting 4n progeny demonstrated cotransmission of Rf3 and rf3 alleles through heterozygous 2n pollen, confirming the dominance of the restoring Rf3 allele (KAMPS et al. 1996 ).

The function of the rfl1 locus:
The homozygous lethality of the rfl1 allele supports the hypothesis that the rfl1 locus is essential to mitochondrial function. This idea is further supported by the observation that the ATPA protein fails to accumulate in rfl1 pollen (Fig 3). Yeast cells that fail to accumulate ATP synthase fail to develop normal mitochondrial structure and function (PAUMARD et al. 2002 ). The mode of rfl1 action, however, is currently unknown. The rfl1 mutation does not affect global mitochondrial protein accumulation because the mitochondrial-encoded COXII protein accumulates to wild-type levels in rfl1 pollen (Fig 3). It is possible, however, that rfl1 affects the accumulation of other mitochondrial-encoded proteins. A role in mitochondrial RNA editing, translation, protein complex assembly, or protein turnover could explain the large decrease in ATPA protein abundance compared to the modest decrease in transcript abundance.

The rfl1 allele might restore male fertility by decreasing the abundance of as-yet-unidentified orf355- and/or orf77-encoded proteins causal to male sterility. Alternatively, the orf355-orf77 gene product or products might require interaction with the ATPA protein to cause male sterility. Yeast mutants that condition a defective ATP synthase with a leaky proton pore can be rescued by second-site mutations that disrupt accumulation of the ATP synthase - or ß-subunits (MUELLER 2000 ). Furthermore, orf77 was recently shown to be modified by plant mitochondrial RNA editing in CMS-S maize microspores. The edited version of orf77 predicts a protein of 17 amino acids having 65% identity to the C-terminal transmembrane domain of ATP synthase subunit 9 (GALLAGHER et al. 2002 ). Expression of the orf17 gene product could potentially result in a leaky proton pore that could be rescued by a mutation disrupting accumulation of ATPA. Further study of rfl1 will provide insights into nuclear regulation of post-transcriptional mitochondrial gene expression, mechanisms of S male sterility, and mechanisms of fertility restoration.

The utility of S cytoplasm in forward genetic analysis of mitochondrial biogenesis:
If the functional allele at the rfl1 locus is required for expression of the mitochondrial CMS locus, a loss-of-function mutation at rfl1 would be readily observed as restoration of fertility in haploid pollen. The CMS-associated orf355-orf77 locus therefore serves as a reporter gene for mitochondrial gene expression. Loss-of-function mutations in nuclear genes necessary for expression of the reporter are observed as male-fertile tassels or tassel sectors, depending upon the developmental timing of the mutation. The gametophytic nature of the CMS-S system affords the recovery of recessive (loss-of-function) restorer-of-fertility mutations that would go unobserved because of dominant masking in the 2n anther cells affected by sporophytic CMS systems.

In addition to allowing for convenient selection of mutations that restore male fertility, the gametophytic CMS-S system also appears to relax the constraints on the types of mutations that are tolerated. Mitochondria increase in number during floral development (CONLEY and HANSON 1995 ), yet rfl1 pollen is functional despite failure to accumulate wild-type levels of ATPA protein. In tobacco, aerobic fermentation occurs during late pollen development and during subsequent pollen germination and tube growth (OP DEN CAMP and KUHLEMEIER 1997 ; TADEGE and KUHLEMEIER 1997 ). The late timing of pollen abortion in the CMS-S system (LEE et al. 1980 ) may be the reason that mutations such as rfl1 are recovered. The recovery of mutations such as rfl1 is difficult or impossible in most obligate aerobes, because mutations disrupting mitochondrial function usually condition lethality or severe developmental abnormalities (NEWTON and GABAY-LAUGHNAN 1998 ; SHOUBRIDGE 2001 ; WALLACE 2001 ). Aerobic fermentation may enable rfl1 pollen to function, whereas rfl1/rfl1 embryos, presumably dependent upon respiration, fail to develop.

Consistent with the nature of the rfl1 mutation and the multiple nuclear loci involved in plant mitochondrial biogenesis (KRUFT et al. 2001 ; MILLAR et al. 2001 ), rfl alleles can be generated at many loci (LAUGHNAN and GABAY 1978 ; GABAY-LAUGHNAN et al. 1995 ). The unique biology and genetics of S male-sterile maize provide, for the first time, a means of systematically selecting nuclear mutations disrupting mitochondrial biogenesis function in a multicellular eukaryote. Because complexity of the mitochondrial proteome is conserved between plants and mammals (EMANUELSSON et al. 2000 ; KARLBERG et al. 2000 ; KRUFT et al. 2001 ), exploitation of the CMS-S system may enhance our understanding of mitochondrial biogenesis and function in a wide range of organisms.

	FOOTNOTES

¹ Present address: Center for Applied Genetic Technologies, University of Georgia, 111 Riverbend Rd., Athens, GA 30602-6810.

	ACKNOWLEDGMENTS

We thank T. Elthon for the ATPA antibody, T. Mason for the COXII antibody, and J. M. Grienenberger for the NAD9 antibody. We thank M. Gallo-Meagher and L. C. Hannah for critically reading the manuscript prior to submission and two anonymous reviewers for their thoughtful suggestions. This work was funded in part by the U.S. Department of Agriculture National Research Initiative competitive grants program award 95-37301-2039 to C.D.C. and award 00-3530-9409 to S.G.-L. and C.D.C. Additional support was provided by the Illinois Foundation Seeds and by the University of Florida Agriculture Experiment Station. This contribution is no. R-09312 in the Florida Agriculture Experiment Station journal series.

Manuscript received January 26, 2003; Accepted for publication May 16, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BENTOLILA, S., A. A. ALFONSO, and M. R. HANSON, 2002  A pentatricopeptide repeat-containing gene restores fertility to cytoplasmic male-sterile plants. Proc. Natl. Acad. Sci. USA 99:10887-10892.[Abstract/Free Full Text]

BUCHERT, J. G., 1961  The stage of the genome-plasmon interaction in the restoration of fertility to cytoplasmically pollen-sterile maize. Proc. Natl. Acad. Sci. USA 47:1436-1440.[Free Full Text]

CHANG, M. T. and M. G. NEUFFER, 1989  Maize microsporogenesis. Genome 32:232-244.

CONLEY, C. S. and M. R. HANSON, 1995  How do alterations in plant mitochondrial genomes disrupt pollen development? J. Bioenerg. Biomembr. 27:447-457.[Medline]

CUI, X., R. P. WISE, and P. S. SCHNABLE, 1996  The rf2 nuclear restorer gene of male-sterile T-cytoplasm maize. Science 272:1334-1336.[Abstract]

DUVICK, D. N., 1965  Cytoplasmic pollen sterility in corn. Adv. Genet. 13:1-56.

EMANUELSSON, O., H. NIELSEN, S. BRUNAK, and G. VON HEIJNE, 2000  Predicting subcellular localization of proteins based on their N-terminal amino acid sequence. J. Mol. Biol. 300:1005-1016.[Medline]

GABAY-LAUGHNAN, S., G. ZABALA and J. R. LAUGHNAN, 1995 S-type cytoplasmic male sterility in maize, pp. 395–432 in Advances in Cellular and Molecular Biology of Plants, Vol. 2: Molecular Biology of the Mitochondria, edited by C. S. LEVINGS, III, and I. K. VASIL. Kluwer, Dordrecht, The Netherlands.

GALLAGHER, L., S. K. BETZ, and C. D. CHASE, 2002  Mitochondrial RNA editing truncates a chimeric open reading frame associated with S male-sterility in maize. Curr. Genet. 42:179-184.[Medline]

GRIVELL, L. A., M. ARTAL-SANZ, G. HAKKAART, L. DE JONG, and L. G. J. NIJTMANS et al., 1999  Mitochondrial assembly in yeast. FEBS Lett. 452:57-60.[Medline]

HANSON, M. R., 1991  Plant mitochondrial mutations and male sterility. Annu. Rev. Genet. 25:461-486.[Medline]

HAYES, W. L., 1973 Statistics for the Social Sciences. Holt, Rinehart and Winston, New York.

KAMPS, T. L. and C. D. CHASE, 1997  RFLP mapping of the maize gametophytic restorer-of-fertility gene, Rf3, and aberrant pollen transmission of the nonrestoring allele, rf3.. Theor. Appl. Genet. 95:525-531.

KAMPS, T. L., D. R. MCCARTY, and C. D. CHASE, 1996  Gametophyte genetics in Zea mays L.: dominance of a restoration of fertility allele (Rf3) in diploid pollen. Genetics 142:1001-1007.[Abstract]

KARLBERG, O., B. CANBÄCK, C. G. KURLAND, and S. G. E. ANDERSSON, 2000  The dual origin of the yeast mitochondrial proteome. Yeast 17:170-187.[Medline]

KRUFT, V., H. EUBEL, L. JÄNSCH, W. WERHAHN, and H.-P. BRAUN, 2001  Proteomic approach to identify novel mitochondrial proteins in Arabidopsis. Plant Physiol. 127:1694-1710.[Abstract/Free Full Text]

LAUGHNAN, J. R. and S. J. GABAY, 1973  Mutations leading to nuclear restoration of fertility in S male-sterile cytoplasm in maize. Theor. Appl. Genet. 43:109-116.

LAUGHNAN, J. R., and S. J. GABAY, 1975 An episomal basis for instability of S male sterility in maize and some implications for plant breeding, pp. 330–349 in Genetics and the Biogenesis of Cell Organelles, edited by C. W. BIRKEY, P. S. PERLMAN and T. J. BYERS. Ohio State University Press, Columbus, OH.

LAUGHNAN, J. R., and S. J. GABAY, 1978 Nuclear and cytoplasmic mutations to fertility in S male-sterile maize, pp. 427–446 in Maize Breeding and Genetics, edited by D. B. WALDEN. John Wiley & Sons, New York.

LEE, S. J., E. D. EARLE, and V. E. GRACEN, 1980  The cytology of pollen abortion in S cytoplasmic male-sterile corn anthers. Am. J. Bot. 67:237-245.

LIU, F., X. CUI, H. T. HORNER, H. WEINER, and P. S. SCHNABLE, 2001  Mitochondrial aldehyde dehydrogenase activity is required for male fertility in maize. Plant Cell 13:1063-1078.[Abstract/Free Full Text]

MILLAR, A., L. J. SWEETLOVE, P. GIEGÉ, and C. J. LEAVER, 2001  Analysis of the Arabidopsis mitochondrial proteome. Plant Physiol. 127:1711-1727.[Abstract/Free Full Text]

MUELLER, D. M., 2000  Partial assembly of the yeast mitochondrial ATP synthase. J. Bioenerg. Biomembr. 32:391-400.[Medline]

NEWTON, K. J., and S. J. GABAY-LAUGHNAN, 1998 Abnormal growth and male sterility associated with mitochondrial DNA rearrangements in plants, pp. 365–381 in Mitochondrial DNA Mutations in Aging, Disease and Cancer, edited by K. K. SINGH. R. G. Landes, Georgetown, TX.

NIVISON, H. T. and M. R. HANSON, 1989  Identification of a mitochondrial protein associated with cytoplasmic male sterility in petunia. Plant Cell 1:1121-1130.[Abstract/Free Full Text]

OP DEN CAMP, R. G. L. and C. KUHLEMEIER, 1997  Aldehyde dehydrogenase in tobacco pollen. Plant Mol. Biol. 35:355-365.[Medline]

PAUMARD, P., J. VAILLIER, B. COULARY, J. SCHAEFFER, and B. SOUBANNIER et al., 2002  The ATP synthase is involved in generating mitochondrial cristae morphology. EMBO J. 21:221-230.[Medline]

RHOADES, M. M. and E. DEMPSEY, 1966  Induction of chromosome doubling at meiosis by the elongate gene in maize. Genetics 54:505-522.[Free Full Text]

SCHNABLE, P. S. and R. P. WISE, 1998  The molecular basis of cytoplasmic male sterility and fertility restoration. Trends Plant Sci. 3:175-180.

SHOUBRIDGE, E. A., 2001  Nuclear genetic defects of oxidative phosphorylation. Hum. Mol. Genet. 8:2277-2284.

SISCO, P. H., V. E. GRACEN, H. L. EVERETT, E. D. EARLE, and D. R. PRING et al., 1985  Fertility restoration and mitochondrial nucleic acids distinguish at least five subgroups among cms-S cytoplasms of maize (Zea mays L.). Theor. Appl. Genet. 71:5-15.

STEINMETZ, L. M., C. SCHARFE, A. M. DEUTSCHBAUER, D. MOKRANJAC, and Z. S. HERMAN et al., 2002  Systematic screen for human disease genes in yeast. Nat. Genet. 31:400-404.[Medline]

STUDENT,, 1908  The probable error of a mean. Biometrika 6:1-25.[Free Full Text]

TADEGE, M. and C. KUHLEMEIER, 1997  Aerobic fermentation during tobacco pollen development. Plant Mol. Biol. 35:343-354.[Medline]

WALLACE, D. C., 2001  A mitochondrial paradigm for degenerative diseases and ageing. Novartis Found. Symp. 235:247-263.[Medline]

WARMKE, H. E. and S.-L. J. LEE, 1977  Mitochondrial degeneration in Texas cytoplasmic male-sterile corn anthers. J. Hered. 68:213-222.[Free Full Text]

WEN, L. and C. D. CHASE, 1999a  Pleiotropic effects of a nuclear restorer-of-fertility locus on mitochondrial transcripts in male-fertile and S male-sterile maize. Curr. Genet. 35:521-526.[Medline]

WEN, L. and C. D. CHASE, 1999b  Mitochondrial gene expression in developing male gametophytes of male-fertile and S male-sterile maize. Sex. Plant Reprod. 11:323-330.

WISE, R. P. and D. R. PRING, 2002  Nuclear-mediated mitochondrial gene regulation and male fertility in higher plants: Light at the end of the tunnel? Proc. Natl. Acad. Sci. USA 99:10240-10242.[Free Full Text]

YATES, F., 1934  Contingency tables involving small numbers and the ² test. J. R. Stat. Soc. 1(Suppl.):217-235.

ZABALA, G., S. GABAY-LAUGHNAN, and J. R. LAUGHNAN, 1997  The nuclear gene Rf3 affects the expression of the mitochondrial chimeric sequence R implicated in S-type male sterility in maize. Genetics 147:847-850.[Abstract]

This article has been cited by other articles:

				H. V. Tang, J. F. Pedersen, C. D. Chase, and D. R. Pring Fertility Restoration of the Sorghum A3 Male-Sterile Cytoplasm through a Sporophytic Mechanism Derived from Sudangrass Crop Sci., May 31, 2007; 47(3): 943 - 950. [Abstract] [Full Text] [PDF]

				M. D. Howell, N. Fahlgren, E. J. Chapman, J. S. Cumbie, C. M. Sullivan, S. A. Givan, K. D. Kasschau, and J. C. Carrington Genome-Wide Analysis of the RNA-DEPENDENT RNA POLYMERASE6/DICER-LIKE4 Pathway in Arabidopsis Reveals Dependency on miRNA- and tasiRNA-Directed Targeting PLANT CELL, March 1, 2007; 19(3): 926 - 942. [Abstract] [Full Text] [PDF]

				Z. Wang, Y. Zou, X. Li, Q. Zhang, L. Chen, H. Wu, D. Su, Y. Chen, J. Guo, D. Luo, 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, March 1, 2006; 18(3): 676 - 687. [Abstract] [Full Text] [PDF]

				M. R. Hanson and S. Bentolila Interactions of Mitochondrial and Nuclear Genes That Affect Male Gametophyte Development PLANT CELL, June 1, 2004; 16(suppl_1): S154 - S169. [Full Text] [PDF]

				S. Gabay-Laughnan, C. D. Chase, V. M. Ortega, and L. Zhao Molecular-Genetic Characterization of CMS-S Restorer-of-Fertility Alleles Identified in Mexican Maize and Teosinte Genetics, February 1, 2004; 166(2): 959 - 970. [Abstract] [Full Text] [PDF]