Cosegregation of Single Genes Associated with Fertility Restoration and Transcript Processing of Sorghum Mitochondrial orf107 and urf209

Corresponding author: Daryl R. Pring, Department of Plant Pathology, USDA-ARS, 1453 Fifield Hall, University of Florida, Gainesville, FL 32611., drpg{at}icbr.ifas.ufl.edu (E-mail).

Communicating editor: K. J. NEWTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Defective nuclear-cytoplasmic interactions leading to aberrant microgametogenesis in sorghum carrying the IS1112C male-sterile cytoplasm occur very late in pollen maturation. Amelioration of this condition, the restoration of pollen viability, involves a novel two-gene gametophytic system, wherein genes designated Rf3 and Rf4 are required for viability of individual gametes. Rf3 is tightly linked to, or represents, a single gene that regulates a transcript processing activity that cleaves transcripts of orf107, a chimeric mitochondrial open reading frame specific to IS1112C. The mitochondrial gene urf209 is also subject to nucleus-specific enhanced transcript processing, 5' to the gene, conferred by a single dominant gene designated Mmt1. Examinations of transcript patterns in F₂ and two backcross populations indicated cosegregation of the augmented orf107 and urf209 processing activities in IS1112C. Several sorghum lines that do not restore fertility or confer orf107 transcript processing do exhibit urf209 transcript processing, indicating that the activities are distinguishable. We conclude that the nuclear gene(s) conferring enhanced orf107 and urf209 processing activities are tightly linked in IS1112C. Alternatively, the similarity in apparent regulatory action of the genes may indicate allelic differences wherein the IS1112C Rf3 allele may differ from alleles of maintainer lines by the capability to regulate both orf107 and urf209 processing activities.

ABNORMAL microsporogenesis or microgametogenesis in higher plants is often associated with deleterious nuclear-cytoplasmic processes resulting in pollen abortion, i.e., cytoplasmic-nuclear male sterility (cms). Many such examples include unique mitochondrial DNA (mtDNA) rearrangements and mutations that are associated with the expression of cms (e.g., LEVINGS and VASIL 1995 ). Other defects result from mtDNA gene mutations that are expressed in somatic cells and show abnormal growth phenotypes, such as nonchromosomal stripe of maize (NEWTON 1995 ). Although somatic abnormalities are apparently not subject to nuclear factors that may reverse these effects, an intriguing facet of cms is that in most cases specific nuclear genes are capable of correcting the deleterious interaction, i.e., the restoration of male fertility. To date only one nuclear fertility restoration gene has been tentatively identified, Rf2 of maize, which may be an aldehyde dehydrogenase (CUI et al. 1996 ).

Differential fertility restoration patterns of male-sterile cytoplasms in sorghum [Sorghum bicolor (L.) Moench] indicate substantial complexity among these cytoplasms and their specific fertility restoration genes. Seven major cms groups have been identified by specific fertility restoration requirements among 22 entries examined (WORSTELL et al. 1984 ; PRING et al. 1995 ; XU et al. 1995 ). The type member of the A3 group, IS1112C, is characterized by the presence of two unique chimeric mtDNA configurations, orf265/130 (TANG et al. 1996A ) and orf107 (TANG et al. 1996B ). The orf107 configuration includes sequences 5' and internal to sorghum atp9 (SALAZAR et al. 1991 ; YAN et al. 1997 ) resulting in a 31-residue amino terminus that is 84% similar to that of ATP9. Uniquely, the carboxy-terminal 49 residues of the predicted orf107 gene product are 80% similar to the predicted carboxy terminus of orf79, an open reading frame of the Chinsurah Boro II male-sterile rice cytoplasm suspected as a cause of cms (IWABUCHI et al. 1993 ; AKAGI et al. 1994 , AKAGI et al. 1995 ).

A highly efficient nucleolytic transcript processing activity (TPA) that cleaves at nucleotide +196 within orf107 was shown to be conferred by a line that restores pollen viability (TANG et al. 1996B ). All progeny partially or fully restored to fertility exhibit enhanced TPA, precluding abundant whole-length transcripts, while all lines that maintain male sterility confer only a trace of the activity. The processing activity results in the cleavage of 1110-, 870-, and 810-nt transcripts concomitant with a marked increase in abundance of a 380-nt transcript, which we conclude is processed since the terminus was not a suitable substrate for guanylyltransferase (YAN and PRING 1997 ). A gametophytic mode of fertility restoration was invoked for the A3 cytoplasm based on absence of male-sterile progeny in F₂-F₄ generations of the cross A3Tx398/IS1112C, or in backcrosses wherein the F₁ was used as male parent (TANG et al. 1996B ). Since each of these restored progeny also exhibits orf107 TPA, we postulated that the TPA was an integral component of fertility restoration.

Nuclear-mitochondrial interactions are often detected as transcript modification through processing, although transcript initiation (NEWTON et al. 1995 ) and editing (LU and HANSON 1992 ; WILSON and HANSON 1996 ) have also been shown to be influenced by nuclear genes. Transcript processing is prominently involved in fertility restoration in several cms examples. The restoration of fertility in T cytoplasm maize requires two genes, Rf1 and Rf2 (e.g., LEVINGS 1993 ). Action of the Rf1 gene results in altered transcripts of the mitochondrial gene T-urf13 (DEWEY et al. 1986 , DEWEY et al. 1987 ; KENNELL and PRING 1989 ). The 5' terminus of the Rf1-conferred transcript is within T-urf13 (DEWEY et al. 1987 ; KENNELL and PRING 1989 ) and the transcript is assumed to be processed since guanylyltransferase capping did not identify a possible initiated RNA species (KENNELL and PRING 1989 ). Rigorous evidence that Rf1 is indeed related to altered T-urf13 transcripts was derived from the generation of rf1-m alleles with the Mutator transposon family, which resulted in male-sterile plants that lost the transcript processing activity (WISE et al. 1996 ).

Fertility restoration of the "pol" male-sterile cytoplasm of Brassica is associated with altered transcript processing 5' to atp6 (SINGH and BROWN 1991 ), which includes sequences of a chimeric open reading frame, orf224. Restoration results from action of either of two dominant loci, Rfp1 or Rfp2, which confers transcript processing of the dicistronic orf224-atp6 transcript, resulting in two transcripts with 5' termini within orf224 (SINGH and BROWN 1991 ). During examinations of this system, SINGH et al. 1996 observed linkage of the Rfp1 gene with another gene, Mmt, that alters transcription of two additional mtDNA genes. They concluded that the recessive rfp1 allele, or the tightly linked gene Mmt, modifies transcripts of nad4 and a gene involved in cytochrome c biogenesis. A unique endonucleolytic cleavage event is associated with fertility restoration of the rice Chinsurah Boro II male-sterile cytoplasm (IWABUCHI et al. 1993 ), possibly releasing a transcript carrying orf79 from a dicistronic atp6-orf79 transcript (IWABUCHI et al. 1993 ; AKAGI et al. 1994 ).

There are several examples of nuclear-conferred transcript processing or other modifications that are not related to the expression of cms in higher plants. Maize "orf25" (DEWEY et al. 1986 ; ROCHEFORD and PRING 1994 ) and the sorghum homolog, urf209 (TANG et al. 1996A ), are subject to nucleus-dependent transcript alteration. Transcripts of urf209 include a 1044-nt transcript and a trace of a 832-nt transcript, characteristic of the lines B3Tx398 and A3Tx398, while a number of other lines confer a marked increase in abundance of the 832-nt transcript (TANG et al. 1996A ). The latter transcript could not be capped with guanylyltransferase, indicating that the transcript results from processing (YAN and PRING 1997 ). Nuclear genes alter transcriptional patterns of atp1 in radish (MAKAROFF and PALMER 1988 ), unrelated to the expression of male sterility (MAKAROFF et al. 1990 ). However, since altered atp1 transcripts showed a correlation with fertility restoration in most nuclear backgrounds, MAKAROFF 1995 suggested that restorer genes and genes affecting other mtDNA transcripts can be linked, as reported for Brassica (SINGH et al. 1996 ).

During investigations to establish the genetic complexity of fertility restoration and the role of transcript processing in male sterility of the A3 sorghum cytoplasm, we observed cosegregation of enhanced orf107 and urf209 transcript processing activities. These activities, or their regulation, are encoded by a single or tightly linked nuclear genes in the line IS1112C. The Rf3-conferred orf107 transcript processing activity is necessary, but not sufficient, to restore male fertility, as part of a unique two-gene gametophytic fertility restoration system.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic stocks:
Sorghum lines used in this study are listed in Table 1. The line IS1112C carries a male-sterile cytoplasm (A3 group; WORSTELL et al. 1984 ), and is fertile by virtue of the presence of requisite nuclear fertility restoration genes. Male-sterile lines A3Tx398 and A3Tx7000 (SCHERTZ et al. 1990 ) carry the IS1112C cytoplasm and are near-isogenic to the male-fertile, normal (N) cytoplasm lines B3Tx398 and B3Tx7000, respectively. The designation B3 indicates that the lines are maintainers of sterility for the A3 cytoplasm. The F₁'s A3Tx398/IS1112C and A3Tx7000/IS1112C were prepared and the former was self-pollinated to generate an F₂ population. The F₁ A3Tx398/IS1112C was used as a male parent to make the backcross A3Tx398//A3Tx398/IS1112C. Since a gametophytic mode of restoration was indicated by previous studies (TANG et al. 1996B ), another backcross was designed to allow an analysis of segregation of the fertility restoration and transcript processing traits. This backcross, A3Tx398/IS1112C//B3Tx398, was made by emasculating A3Tx398/IS1112C and pollinating with B3Tx398. The F₁ B3Tx398/IS1112C was also made for inheritance studies.

IS1112C and all derived lines exhibit strong tillering and aerial branching traits, allowing near-immortalization of important stocks. Progeny plants grown in the greenhouse were set out in plots and, following phenotype scoring and selfing/crossing, cut back and transplanted into pots and placed in the greenhouse.

Pollen ratings and fertility determinations:
Pollen maturation was scored from field-grown plants by stainability with a 1% iodine-1% potassium iodide solution. Several florets from each plant were collected prior to anthesis and typically three anthers were excised and disrupted, and pollen was scored by microscopic examination. Several hundred pollen grains from each anther were scored as fully stained, partially stained, or nonstained. Plants were scored as nonrestored or partially to fully restored, in terms of pollen stainability. Pollen from B3Tx398 plants was about 95% stainable. For pollen diameter measurements, freshly harvested anthers were immediately disrupted, stained, and photographed. Measurements of pollen diameter were made from enlargements, using a stage micrometer for calibration. For some measurements pollen was stained with a 1:5 dilution of iodine-potassium iodide. Fertility was evaluated by seed set in panicles that were bagged upon emergence, and self-pollinated. Nonrestored plants are sterile, while partially to fully restored plants are fertile and set seed.

RNA preparation and analysis:
mtRNAs from sorghum leaves of greenhouse-/field-grown plants were isolated by methods previously described (TANG et al. 1996A , TANG et al. 1996B ), or by an abbreviated procedure. Typically, 5-g leaf samples were processed, subjected to one cycle of differential centrifugation, lysed, and subjected to LiCl precipitation. Northern analyses were conducted with RNA equivalent to 1–2 g fresh tissue. The probes used for Northern analyses were pHT160, an urf209 cDNA clone (TANG et al. 1996A ), and the orf107 clone pHC104, which includes shared atp9-orf107 sequences, allowing a comparison of orf107 transcript abundance relative to that of atp9 (TANG et al. 1996B ). Hybridization procedures were as previously described (TANG et al. 1996A ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fertility restoration is by a two-gene gametophytic system wherein one gene confers or regulates orf107 transcript processing:
The modification of orf107 transcripts by processing was postulated to be necessary but not sufficient for fertility restoration of the A3 source of cms (TANG et al. 1996B ), based on presence of the activity in all partially or fully restored progeny. In these tests, 10 partially restored plants from the F₁ A3Tx398/IS1112C, 29 partially or fully restored F₂'s from A3Tx398/IS1112C, and 20 partially restored plants from the backcross A3Tx398//A3Tx398/IS1112C each exhibited TPA (Table 2). Processing of orf107 transcripts in these lines is manifested as a 75% decrease in abundance of 1100-, 870-, and 810-nt whole-length transcripts compared to TPA(-) lines (Figure 1A), as determined by imaging and densitometry, and marked accumulation of the 3' processed 380-nt product. Importantly, a trace basal level of the processed 380-nt transcript in all male-sterile lines that carry the A3 cytoplasm (TANG et al. 1996B ) indicates the possibility that processing is enhanced in TPA(+) lines.

View larger version (59K): In this window In a new window Download PPT slide   Figure 1. Analyses of orf107 and urf209 transcriptional patterns indicate that transcript processing conferred by nuclear genes may be regulatory. (A) Orf107 transcripts of 1110-, 870-, 810-, and 380-nt and a 650-nt atp9 transcript in Rf3Rf3 [orf107 TPA(+)] and rf3rf3 [orf107 TPA(-)] lines probed with the clone pHC104 (TANG et al. 1996B ). The probe includes sequences of atp9, allowing a comparison of abundance of the 650-nt atp9 transcript relative to those of orf107. (B) Urf209 transcripts of 1044 and 832 nt in Mmt1Mmt1 [urf209 TPA(+)] and mmt1mmt1 [urf209 TPA(-)] lines probed with an urf209 cDNA clone (TANG et al. 1996A ). The autoradiogram was overexposed to illustrate the 832-nt transcript in mmt1mmt1 lines. Arrows indicate presence of trace levels of the processed 380-nt orf107 and 832-nt urf209 transcripts in "-" lines.

The absence of sterile plants among the backcross and the F₂ progeny is consistent with a gametophytic mode of restoration, where only gametes carrying requisite restorer alleles will be viable, and thus only restorer alleles are transmitted to progeny through the male parent. These progeny, however, do not allow an analysis of genetic complexity of the presumed gametophytic mode of restoration. To determine the complexity of restoration, the fertile F₁ A3Tx398/IS1112C was emasculated and pollinated with the maintainer line B3Tx398, generating the backcross line A3Tx398/IS1112C//B3Tx398. Individuals from this backcross segregated for fertile and sterile plants, and 81 plants were examined for orf107 TPA and field-grown for fertility assessments. An example of 21 such progeny shows 11 with a trace level of processing and abundant 1110-, 870-, and 810-nt transcripts (Figure 2A, "-"). The remaining 10 individuals with enhanced orf107 transcript processing show reduction in abundance of the precursor transcripts and a concomitant increase in abundance of the processed, 380-nt transcript (Figure 2A, "+").

View larger version (51K): In this window In a new window Download PPT slide   Figure 2. Enhanced processing of orf107 and urf209 transcripts cosegregates in 21 progeny of the backcross A3Tx398/IS1112C//B3Tx398. (A) Transcript patterns of progeny probed with orf107 clone pHT104. (B) Transcript patterns of progeny probed with an orf209 cDNA clone. Note cosegregation of appearance of increased 380-nt orf107 and 832-nt urf209 transcripts, noted by +.

The 81 field-grown progeny segregated 37 TPA(+):44 TPA(-), indicating that a single nuclear gene confers the enhanced processing activity (Table 2). Chi square analysis shows lack of concordance with a two-gene model (1:3; ² = 18.47). Fertility observations of the 81 progeny indicated that all 44 plants with negligible processing activity were sterile, consistent with the assumption that TPA is required for restoration. Among the 37 plants with processing activity, however, 16 were sterile. The latter exceptional plants, TPA(+) sterile, indicate that an additional gene(s), in addition to the gene influencing TPA, is required for fertility restoration. Segregration of fertile:sterile plants (21:60) fits a 1:3 ratio (Table 2), indicating that fertility restoration involves two genes. Chi square tests show a lack of concordance with either a one- (expected ratio of 1:1) or three-gene (expected ratio of 1:7) model (² = 15.73 and 13.35, respectively). A two-gene model involving orf107 TPA as one component should also fit an expected ratio of 1 [TPA(+), fertile]:1 [TPA(+), sterile]:2 [TPA(-), sterile]. The observed frequency of the 81 plants examined, 21:16:44, fits this model (Table 2).

We have designated the gene influencing orf107 processing activity Rf3, and the second gene, Rf4, in consultation with Drs. K. F. SCHERTZ and J. E. MULLET, Texas A&M University, College Station, Texas (personal communication). Since both genes are required for gamete viability, gene action is complementary. Dominance cannot readily be assigned in gametophytic restoration systems and our designations of the dominant alleles is thus tentative. The presumed genotypes Rf3Rf3Rf4Rf4 can be assigned to IS1112C, and rf3rf3rf4rf4 to the maintainer line B3Tx398. According to this two-gene model, the 44 TPA(-) sterile plants from the backcross A3Tx398/IS1112C//B3Tx398 are rf3rf3rf4rf4 or heterozygous for Rf4, i.e., rf3rf3Rf4rf4. The 21 TPA(+) fertile individuals are Rf3rf3Rf4rf4, like the F₁ A3Tx398/IS1112C, while the 16 sterile plants in this population are Rf3rf3rf4rf4.

The expression of cms and fertility restoration is consistent with that of a gametophytic system:
The manifestation of cms in gametophytic restoration systems occurs postmeiotically, during microgametogenesis. Pollen abortion in S cytoplasm maize, for instance, is characterized by degradation and collapse following intine deposition, very late in pollen maturation (LEE et al. 1980 ). Therefore, we examined pollen development in the sorghum lines. Male-sterile A3Tx398 plants exsert bright yellow-orange anthers indistinguishable from those of B3Tx398, in a cursory visual examination. In contrast to many other cms examples wherein aborted microspores collapse and are visualized as shrunken pollen at what would be anthesis, anthers of male-sterile plants examined within 24 hr of anthesis were engorged with turgid, nearly mature pollen (Figure 3; Table 3). Measurements of pollen diameter, and iodine-potassium iodide staining, illustrate the late stage at which pollen development was arrested. At about 8 days prior to anthesis, when the panicle has not yet emerged, the diameter of pollen from A3Tx398 and B3Tx398 plants (35–36 µm) was identical. About 24 hr prior to anthesis, the diameter of pollen from A3Tx398 had increased 32% to 47 µm, while that of B3Tx398 had increased 48% to 53 µm. Consequently, the diameter and volume of sterile pollen were 89 and 70%, respectively, of that of fertile pollen (Table 3). Staining of these pollen samples showed that fertile and sterile pollen grains stain the same brown-red with iodine-potassium iodide at 8 days prior to anthesis (Figure 3A and Figure B), whereas at 24 hr prior to anthesis sterile pollen was unchanged in color and fertile pollen had acquired the typical dark brown-black color indicative of amylose formation (Figure 3C and Figure D). Parallel examinations of the F₁ A3Tx398/IS1112C at 1 day prior to anthesis, with assignment to nonstaining or stained categories, revealed the same diameter categories as A3Tx398 and B3Tx398 (Table 3). We conclude that the maturation of fertile pollen includes a dramatic increase in volume and the appearance of detectable amylose within 8 days of anthesis. It is at this very late stage that the maturation of sterile pollen is arrested.

View larger version (121K): In this window In a new window Download PPT slide   Figure 3. Sorghum pollen from IS1112C (A and B) and A3Tx398 (C and D) and at 8 (A and C) and 1 (B and D) day prior to anthesis, stained with 1% iodine-1% potassium iodide. x136. (E) Pollen from the F1 A3Tx398/IS1112C 1 day prior to anthesis stained with 0.2% iodine-0.2% potassium iodide and examined under phase-contrast microscopy. Note partial amylose deposition in some of the pollen grains.

Examination of pollen stainability in the F₁ A3Tx398/IS1112C 1 day prior to anthesis revealed a gradation of pollen stainability types. The sterile grains stained brown-red, as in A3Tx398, and many fertile grains stained dark brown-black, as in B3Tx398. Visually intermediate classes were also evident wherein amylose synthesis proceeded such that part of the grain stained black. These grains were better visualized by staining with the 1:5 dilution of the standard iodine-potassium iodide stain and examination by phase-contrast microscopy (Figure 3E). We designated these partially staining pollen grains as partials and included these in a tentative "fertile" category, since these partially stainable grains do not imic sterile pollen in A3Tx398.

A two-gene gametophytic restoration system should yield 25% viable pollen grains in the F₁ A3Tx398/IS1112C, assuming the two parents are rf3rf3rf4rf4 and Rf3Rf3Rf4Rf4. Inspection and visual assignment of iodine stainability of pollen sampled from 10 such F₁'s revealed 4.7% fully stained pollen and 12.5% intermediate grains, giving 17.2% that were "stainable" to some degree. Similarly, examinations of nine A3Tx7000/IS1112C progeny gave 5.1% fully stained pollen and 12.5% intermediate, and thus 17.7% stainable pollen. Under field conditions B3Tx398 pollen was about 95% stainable; thus within the constraints of assigning stainability the F₁ data are reasonably close to 25%. Examinations of pollen stainability in a number of F₂'s showed segregation for discrete categories of ca. 20–25%, 45–50%, and essentially totally stained pollen, consistent with the postulated two-gene system.

The Rf3 restorer allele is linked to Mmt1, a single dominant gene conferring urf209 transcript modification:
Transcription of the mitochondrial gene urf209 in B3Tx398 or A3Tx398 is characterized by a major initiated 1044-nt transcript, and a trace of a processed 832-nt transcript (Figure 1B; TANG et al. 1996A ). IS1112C and other lines confer a transcript processing event resulting in an apparent reduction in abundance of the 1044-nt transcript and an increase in the 832-nt species (TANG et al. 1996A ). We examined the genetic complexity of urf209 transcript processing in crosses of A3Tx398 and IS1112C. The Northern blots utilized in extensive orf107 transcript studies were used to examine urf209 transcript processing. These blots included 10 individual A3Tx398/IS1112C F₁'s, 29 F₂'s, and 81 A3Tx398/IS1112C//B3Tx398 backcross plants that had been scored in the field for fertility ratings (Table 2).

Unexpectedly, we observed complete coinheritance of orf107 and urf209 transcript processing in these progeny. Each of 10 F₁'s and 29 F₂'s exhibited orf107 and urf209 TPA. Most importantly, analyses of progeny of the backcross A3Tx398/IS1112C//B3Tx398 showed that each plant exhibiting orf107 TPA also showed urf209 TPA (Figure 2B). The 37 progeny of this backcross that were orf107 TPA(+) (Table 2) were also urf209 TPA(+), and the 44 sterile orf107 TPA(-) plants were orf209 TPA(-). Thus the ² values for segregation of orf107 TPA are applicable to urf209 TPA. These observations establish that urf209 TPA is conferred by a single gene and has a gametophytic mode of inheritance in the A3 cytoplasm because of linkage to Rf3.

Since the male-sterile line A3Tx7000 and the maintainer line B3Tx7000 both exhibit urf209 TPA (TANG et al. 1996A ), a test was devised to determine if the genetic complexity conferred by this line is similar to that of IS1112C. Male-sterile progeny of the reciprocal crosses A3Tx398/B3Tx7000 and A3Tx7000/B3Tx398 each exhibited urf209 TPA, indicating that presence of the processed transcript is dominant. The male-sterile F₁'s were backcrossed as female to give lines A3Tx7000/B3Tx398//B3Tx7000 and A3Tx398/B3Tx7000//B3Tx398, and the progeny were examined. Consistent with an assumption that B3Tx7000 carries the dominant allele, 17 progeny of the A3Tx7000/B3Tx398//B3Tx7000 each exhibited urf209 TPA (Table 4). The progeny of A3Tx398/B3Tx7000//B3Tx398 segregated 22 TPA(+):14 TPA(-). Chi-square analyses of the latter progeny (Table 4) indicate that the ratio fits both one- and two-gene models. Thus, an unambiguous determination of one or two genes cannot be made for B3Tx7000 with the population size utilized.

Neither of the above tests enable a determination of dominance of the functional allele conferring urf209 TPA through F₂ analyses. Progeny of crosses involving A3Tx398 as female do not transmit the recessive allele for orf107 TPA, which is linked to the single gene conferring urf209 TPA, and the progeny of the B3Tx398 and B3Tx7000 crosses were male-sterile. Therefore we made the cross B3Tx398/IS1112C to allow transmission of all alleles through both parents. Ten F₁ plants each showed urf209 TPA and the F₂ segregated 32 TPA(+):11 TPA(-) (Table 4). These data are consistent with a conclusion that the allele conferring processing is dominant. In this particular population ² analyses showed that segregation in the F₂ fits a single gene model (3:1) better than a two-gene model (13:3), although the values allow acceptance of both hypotheses. Cumulatively we conclude that urf209 TPA is controlled by a single dominant gene in IS1112C. We have designated this gene as Mmt1 (modifier of mitochondrial transcripts), following the precedent of Mmt in Brassica (SINGH et al. 1996 ).

We have also observed urf209 transcript processing in the lines 3-Dwarf White Sooner Milo (3-Dwarf Milo) and IS12662C (data not shown), which are the sources of the A1 and A2 male-sterile cytoplasms, respectively (WORSTELL et al. 1984 ), and do not restore fertility to A3Tx398 (WORSTELL et al. 1984 ; TANG et al. 1996B ). As expected, the lines A1Tx398 and A2Tx398 have the genotype mmt1mmt1. The line 3-Dwarf Milo conferred urf209 TPA in the restored F₁ A1Tx398/3-Dwarf Milo and the male-sterile F₁ A3Tx398/3-Dwarf Milo. Similarly IS12662C confers urf209 TPA in the restored F₁ A2Tx398/IS12662C. B3Tx398 is the only line examined to date that is mmt1mmt1. This line may therefore be unique, and interestingly, is also a maintainer for all currently known sorghum male-sterile cytoplasms (WORSTELL et al. 1984 ; XU et al. 1995 ).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The restoration of male fertility in lines carrying the A3 source of cms in sorghum is gametophytic and uniquely dependent on the action of two genes, designated Rf3 and Rf4, each required for maturation of individual gametes. Cumulatively, we have observed only one male-sterile plant among nearly 450 F₂-F₄ and backcross progeny wherein partially restored plants were used to pollinate A3Tx398 (TANG et al. 1996B ). The absence of male-sterile segregants in the A3Tx398//A3Tx398/IS1112C backcross, together with the segregation ratios observed only in the A3Tx398/IS1112C//B3Tx398 backcross, indicates a gametophytic mode of restoration. All viable gametes carry the restorer alleles, and thus the male parent in the A3 cytoplasm does not transmit nonrestoring alleles. Similarly, enhanced processing of orf107 transcripts was evident in all such progeny. The presence of orf107 TPA in 90 A3Tx398//A3Tx398/IS1112C backcross progeny and 29 F₂'s from A3Tx398/IS1112C (TANG et al. 1996B ) is consistent with gametophytic inheritance and a role of transcript processing in fertility restoration. To our knowledge a two-gene gametophytic fertility restoration system has not been previously documented in higher plants. Single-gene gametophytic restoration systems include S cytoplasm maize involving the dominant Rf3 allele (GABAY-LAUGHNAN et al. 1995 ; KAMPS et al. 1996 ), and the rice Chinsurah Boro II cytoplasm, which requires a gene designated Rf1 (IWABUCHI et al. 1993 ).

The manifestion of pollen abortion in the A3 sorghum cytoplasm exhibits characteristics of aberrant microgametogenesis, consistent with a gametophytic restoration pattern. The unusual feature of near-mature pollen grains in exserted anthers of sterile plants indicates that aberrant processes leading to pollen abortion occur very late in pollen maturation. Cytological investigations of pollen biogenesis in sorghum indicate that intine deposition occurs at the vacuolate pollen stage, when anthers are about 75–90% of full length, and tapetal cells have collapsed (CHRISTENSEN et al. 1972 ). Pollen from A3Tx398 or nonstained pollen from the F₁ A3Tx398/IS1112C clearly reaches this late stage prior to abortion. In contrast, degeneration of pollen in S cytoplasm maize occurs abruptly following intine deposition and before disappearance of the central vacuole (LEE et al. 1980 ), and anthers are not normally exserted (GABAY-LAUGHNAN et al. 1995 ). Amylose deposition in pollen of male-fertile sorghum begins within 8 days of anthesis, and absence of detectable amylose in pollen from sterile plants clearly is an indicator of malfunction prior to or during this late stage of pollen maturation. The variation in amylose deposition among the population of apparently fertile pollen grains in the partially restored F₁ also suggests a gradation in restitution of normal processes leading to gamete viability. An indication that nonstained pollen has lost integrity is reflected in observations that these pollen grains desiccate and collapse within a few hours following anther exsertion.

Linkage of fertility restoration and enhanced transcript processing of orf107 indicates that Rf3 may be regarded as conferring or regulating transcript processing. Cosegregation of Rf3 and the urf209-transcript modifier Mmt1 in IS1112C represents a second example of linkage of nuclear genes that restore male fertility and modify expression of other mtDNA genes. SINGH et al. 1996 demonstrated that the Rpf1 gene of B. napus cosegregates with processing of an orf224-atp6 dicistronic transcript, which cleaves the cms-associated orf224 internally. The recessive rfp1, or the tightly linked Mmt, acts as a dominant gene in processing transcripts of two other mitochondrial gene regions. In the sorghum example we can assign dominance to the Mmt1 allele, but not to the linked Rf3. A determination of the functional allele of Rf3 will require the use of tetraploid stocks, as utilized to assign dominance to the maize Rf3 allele (KAMPS et al. 1996 ). Since Mmt1 is linked to Rf3 in IS1112C, we cannot reject the possibility that Mmt1 plays a role in the fertility restoration. Rf3 and Rf4 are both required for gamete viability, and each viable pollen grain thus is Mmt1 Rf3 Rf4. Since complementary gene action is indicated, the presence of Mmt1 in male-sterile lines such as A3Tx7000 (rf3rf3, Mmt1Mmt1) is not unexpected. The restoring locus could consist of Rf3 and Mmt1, and recombination and recovery of Rf3/mmt1 segregants, with subsequent reproducibility of fertility restoration, would be required to assess a possible role of Mmt1 in restoration.

The observation of a trace of the orf107 and urf209 processed transcripts in all lines examined to date raises the distinct possibility that Rf3 and Mmt1 may be regulatory genes. An alternative explanation to the Rf3-Mmt1 linkage is allelic variation among sorghum lines. Under this hypothesis, the sorghum activities may have derived from a single gene. The IS1112C allele [Rf3, orf107 TPA(+), urf209 TPA(+)] may differ from the maintainer alleles of most lines [rf3, orf107 TPA(-), urf209 TPA(+)] by the additional capability to confer the activity on the orf107 template. The line B3Tx398 [rf3, orf107 TPA(-), urf209 TPA(-)] may be characterized by an allele that confers neither activity. SINGH et al. 1996 invoked allelic variation in the evolution of Rfp1, suggesting that the allele may have originated from a gene influencing normal transcript modification processes. Allelic variation as a basis of fertility restoration may have a precedent in the identification of the maize restorer gene Rf2, which probably encodes an aldehyde dehydrogenase (CUI et al. 1996 ). Interestingly, a line carrying the reference rf2 allele displayed a transcript the same size as that of Rf2 lines (CUI et al. 1996 ), and polyclonal antibody to the aldehyde dehydrogenase recognized a mitochondrial protein in the rf2 line (P. S. SCHNABLE, personal communication). In this case the recessive allele may be inadequate to fullfill the Rf2 function in diploid tapetal cells or microspore mother cells. The putative sorghum Rf3-Mmt1 allelic relationship could have evolved from a component of an RNA processing mechanism.

Transcript processing similar to that affecting orf107 has been invoked for action of the fertility restoration genes Rf1 in maize (DEWEY et al. 1987 ; KENNELL and PRING 1989 ) and Rpf1 and Rfp2 in Brassica (SINGH and BROWN 1991 ; SINGH et al. 1996 ). The efficiency of the Rf3-conferred processing activity reported here, 75% cleavage of progenitor transcripts, is clearly more apparent than in the Rf1-maize and Rpf1/Rpf2-Brassica examples. Interestingly, there are elements of sequence similarity among the maize and sorghum processing sites. Sequences overlapping the 5' termini of maize T-urf13 transcripts processed by action of the maize fertility restorer gene Rf1 or two additional partial fertility restorers, Rf8 and Rf*, share significant similarity with sequences overlapping the 5' terminus of the sorghum Rf3-conferred 380-nt transcript (DILL et al. 1997 ). A conserved motif 5'-CNACNNU-3' was derived for this site, where the sorghum sequence is CUACAAU; within the members of the motif group the Rf*-conferred site shares 8 of 9 bp with the Rf3 recognition site: AC(C/U)ACAAUA. We were unable to identify a similar motif near the Mmt1-conferred processing site 5' to urf209 (TANG et al. 1996A ). The conserved orf107 and T-urf13 processing motifs differ from the conserved sequences associated with Rfp1/Mmt processing in Brassica (SINGH et al. 1996 ). Additionally, in contrast to the maize and sorghum examples the Brassica sequences are located several bp 3' to the processing site, within the processed transcripts.

Little data are available on mechanisms of 5' transcript processing in higher plant mitochondria. Substantial progress has been made in elucidating chloroplast 3' RNA processing, which involves complex interactions including exo- and endoribonucleases (e.g., HAYES et al. 1996 ; YANG et al. 1996 ). Sequence-specific cleavage has been demonstrated for 5' and 3' mRNA processing in yeast mitochondria. Processing 5' can involve cleavage 5' to a conserved sequence and the site-specific endonuclease, RNase MRP (STOHL and CLAYTON 1992 ), and 3' processing is associated with cleavage 3' to a common dodecamer sequence, which may function in pre- and mature mRNAs (HOFMANN et al. 1993 ). In addition, regulation of 5' processing has been shown to be conferred by the yeast nuclear gene PET127, which plays a role in RNA surveillance and/or processing, and is necessary for efficient 5' processing of certain mitochondrial gene transcripts (WIESENBERGER and FOX 1997 ). Mutants of PET127 had no effect on some gene transcripts, but allowed accumulation of unprocessed precursors of transcripts of four genes, and altered transcripts of two other genes, suggesting a pleiotropic role in regulating general processing. The apparent regulation of orf107 and urf209 processing suggests the possibility of a PET127-like system and variants of regulatory factors that lead to altered processing efficiency.

In the sorghum orf107, maize (DILL et al. 1997 ) and Brassica (SINGH et al. 1996 ) examples, accumulation of the 3' products resulting from processing clearly indicates stabilization of the 3' product and destabilization of the 5' product. The higher plant examples are also characterized by the presence of conserved sequences near 5' termini of the processed transcripts, albeit variation between the sorghum-maize and Brassica sequences, which provide ample evidence of site-specific recognition factors involved in transcript processing. In the case of orf107, the target cleavage site may have been ancestrally positioned 5' to a normal gene elsewhere in the genome and recombinational events positioned the site within the chimeric orf107. Selection pressures would then have ostensibly allowed coevolution of a cms factor and the corresponding nuclear mechanism to prevent expression of the trait.

	ACKNOWLEDGMENTS

We thank Drs. C. D. CHASE, P. S. CHOUREY, and D. W. GABRIEL for critically reviewing the manuscript, and Drs. P. S. SCHNABLE and R. P. WISE for sharing unpublished data. The continued counsel of Dr. K. F. SCHERTZ, U.S. Department of Agriculture-Agricultural Research Service, retired, is gratefully acknowledged. This research was supported in part by the USDA-ARS Research Associate program, Cooperative Investigations, USDA-ARS, and Florida Agricultural Experiment Station, Institute of Food and Agricultural Sciences, University of Florida. This is part of Journal Series number R-05927, Institute of Food and Agricultural Sciences, University of Florida.

Manuscript received November 3, 1997; Accepted for publication May 14, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AKAGI, H., M. SAKAMOTO, C. SHINJYO, H. SHIMADA, and T. FUJIMURA, 1994  A unique sequence located downstream from the rice mitochondrial atp6 may cause male sterility. Curr. Genet. 25:52-58[Medline].

AKAGI, H., A. NAKAMURA, R. SAWADA, M. OKA, and T. FUJIMURA, 1995  Genetic diagnosis of cytoplasmic male sterile cybrid plants of rice. Theor. Appl. Genet. 90:948-951.

CHRISTENSEN, J. E., H. T. HORNER, JR., and N. R. LERSTEN, 1972  Pollen wall and tapetal orbicular wall development in Sorghum bicolor (Gramineae). Am. J. Bot. 59:43-58.

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].

DEWEY, R. E., C. S. LEVINGS, III, and D. H. TIMOTHY, 1986  Novel recombinations in the maize mitochondrial genome produce a unique transcriptional unit in the Texas male-sterile cytoplasm. Cell 44:439-449[Medline].

DEWEY, R. E., D. H. TIMOTHY, and C. S. LEVINGS, III, 1987  A unique mitochondrial protein associated with cytoplasmic male sterility in the T cytoplasm of maize. Proc. Natl. Acad. Sci. USA 84:5374-5378[Abstract/Free Full Text].

DILL, C. L., R. P. WISE, and P. S. SCHNABLE, 1997  Rf8 and Rf* mediate unique T-urf13-transcript accumulation, revealing a conserved motif associated with RNA processing and restoration of pollen fertility restoration in T-cytoplasm maize. Genetics 147:1367-1379[Abstract].

GABAY-LAUGHNAN, S., G. ZABALA and J. R. LAUGHNAN, 1995 S-type cytoplasmic male sterility in maize, pp. 395–432 in The Molecular Biology of Plant Mitochondria, edited by C. S. LEVINGS, III, and I. K. VASIL. Kluwer Academic Publishers, Boston.

HAYES, R., J. KUDLA, G. SCHUSTER, L. GABAY, and P. MALIGA et al., 1996  Chloroplast mRNA 3'-end processing by a high molecular weight protein complex is regulated by nuclear encoded RNA binding proteins. EMBO J. 15:1132-1141[Medline].

HOFMANN, T. J., J. MIN, and H. P. ZASSENHAUS, 1993  Formation of the 3' end of yeast mitochondrial mRNAs occurs by site-specific cleavage two bases downstream of a conserved dodecamer sequence. Yeast 9:1319-1330[Medline].

IWABUCHI, M., J. KYOZUKA, and K. SHIMAMOTO, 1993  Processing followed by complete editing of an altered mitochondrial atp6 RNA restores fertility of cytoplasmic male sterile rice. EMBO J. 12:1437-1446[Medline].

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

KENNELL, J. C. and D. R. PRING, 1989  Initiation and processing of atp6, T-urf13 and ORF221 transcripts from mitochondria of T cytoplasm maize. Mol. Gen. Genet. 216:16-24.

LEE, S.-L. 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.

LEVINGS, C. S., III, 1993  Thoughts on cytoplasmic male sterility in cms-T maize. Plant Cell 5:1285-1290[Free Full Text].

LEVINGS, C. S., III, and I. K. VASIL (Editors), 1995 The Molecular Biology of Plant Mitochondria. Kluwer Academic Publishers, Boston.

LU, B. and M. R. HANSON, 1992  A single nuclear gene specifies the abundance and extent of RNA editing of a plant mitochondrial transcript. Nucleic Acids Res. 20:5699-5703[Abstract/Free Full Text].

MAKAROFF, C. A., 1995 Cytoplasmic male sterility in Brassica species, pp. 515–555 in: The Molecular Biology of Plant Mitochondria, edited by C. S. LEVINGS, III and I. K. VASIL. Kluwer Academic Publishers, Boston.

MAKAROFF, C. A. and J. D. PALMER, 1988  Mitochondrial DNA rearrangements and transcriptional alterations in the male-sterile cytoplasm of Ogura radish. Mol. Cell. Biol. 8:1474-1490[Abstract/Free Full Text].

MAKAROFF, C. A., I. J. APEL, and J. D. PALMER, 1990  Characterization of radish mitochondrial atpA: influence of nuclear background on transcription of atpA-associated sequences and relationship with male sterility. Plant Mol. Biol. 15:735-746[Medline].

NEWTON, K. J., 1995 Aberrant growth phenotypes associated with mitochondrial genome rearrangements in higher plants, pp. 585–596 in The Molecular Biology of Plant Mitochondria, edited by C. S. LEVINGS, III and I. K. VASEL. Kluwer Academic Publishers, Boston.

NEWTON, K. J., B. WINBERG, K. YAMATO, S. LUPOLD, and D. B. STERN, 1995  Evidence for a novel mitochondrial promoter preceding the cox2 gene of perennial teosintes. EMBO J. 14:585-593[Medline].

PRING, D. R., H. V. TANG and K. F. SCHERTZ, 1995 Cytoplasmic male sterility and organelle DNAs of sorghum, pp. 461–495 in The Molecular Biology of Plant Mitochondria, edited by C. S. LEVINGS, III and I. K. VASIL. Kluwer Academic Publishers, Boston.

ROCHEFORD, T. R. and D. R. PRING, 1994  Interactions of nuclear and mitochondrial genomes in the alteration of maize mitochondrial orf221 transcripts. Theor. Appl. Gen. 89:951-958.

SALAZAR, R. A., D. R. PRING, and F. KEMPKEN, 1991  Editing of mitochondrial atp9 transcripts from two sorghum lines. Curr. Genet. 20:483-486[Medline].

SCHERTZ, K. F., L. E. CLARK, and D. T. ROSENOW, 1990  Registration of A3Tx430 and A3Tx7000 sorghum genetic stocks. Crop Sci. 30:967[Abstract/Free Full Text].

SINGH, M. and G. G. BROWN, 1991  Suppression of cytoplasmic male sterility by nuclear genes alters expression of a novel mitochondrial gene region. Plant Cell 3:1349-1362[Abstract/Free Full Text].

SINGH, M., N. HAMEL, R. MENASSA, X.-Q. LI, and B. YOUNG et al., 1996  Nuclear genes associated with a single Brassica CMS restorer locus influence transcripts of three different mitochondrial gene regions. Genetics 143:505-516[Abstract].

STOHL, L. L. and D. A. CLAYTON, 1992  Saccharomyces cerevisiae contains an RNase MRP that cleaves at a conserved mitochondrial RNA sequence implicated in replication priming. Mol. Cell. Biol. 12:2561-2569[Abstract/Free Full Text].

TANG, H. V., D. R. PRING, F. R. MUZA, and B. YAN, 1996a  Sorghum mitochondrial orf25 and a related chimeric configuration of a male-sterile cytoplasm. Curr. Genet. 29:265-274[Medline].

TANG, H. V., D. R. PRING, L. C. SHAW, F. A. SALAZAR, and F. A. MUZA et al., 1996b  Transcript processing internal to a mitochondrial open reading frame is correlated with fertility restoration in male-sterile sorghum. Plant J. 10:123-133[Medline].

WIESENBERGER, G. and T. D. FOX, 1997  Pet127p, a membrane-associated protein involved in stability and processing of Saccharomyces cerevisiae mitochondrial RNAs. Mol. Cell. Biol. 17:2816-2824[Abstract].

WILSON, R. K. and M. R. HANSON, 1996  Preferential RNA editing at specific sites within transcripts of two plant mitochondrial genes does not depend on transcriptional context or nuclear genotype. Curr. Genet. 30:502-508[Medline].

WISE, R. P., C. L. DILL, and P. S. SCHNABLE, 1996  Mutator-induced mutations of the rf1 nuclear fertility restorer of T-cytoplasm maize alter the accumulation of T-urf13 mitochondrial transcripts. Genetics 143:1383-1394[Abstract].

WORSTELL, J. V., H. J. KIDD, and K. F. SCHERTZ, 1984  Relationships among male-sterility inducing cytoplasms of sorghum. Crop Sci. 24:186-189[Abstract/Free Full Text].

XU, G.-W., Y.-X. CUI, K. F. SCHERTZ, and G. E. HART, 1995  Isolation of mitochondrial DNA sequences that distinguish male-sterility-inducing cytoplasms in Sorghum bicolor (L.) Moench. Theor. Appl. Genet. 90:1180-1187.

YAN, B. and D. R. PRING, 1997  Transcriptional initiation sites in sorghum mitochondrial DNA indicate conserved and variable features. Curr. Genet. 32:287-295[Medline].

YAN, B., R. A. SALAZAR, and D. R. PRING, 1997  Transcription initiation sites are positioned immediately 3' to trnfM in sorghum mitochondrial DNA. Plant Mol. Biol. 34:823-829[Medline].

YANG, J., G. SCHUSTER, and D. B. STERN, 1996  CSP41, a sequence-specific chloroplast mRNA binding protein, is an endoribonuclease. Plant Cell 8:1409-1420[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]

				J. Zhang, J. M. Stewart, and T. Wang Linkage Analysis between Gametophytic Restorer Rf2 Gene and Genetic Markers in Cotton Crop Sci., January 1, 2005; 45(1): 147 - 156. [Abstract] [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]

				H. V. Tang and D. R. Pring Conversion of Fertility Restoration of the Sorghum IS1112C (A3) Male-Sterile Cytoplasm from Two Genes to One Gene Crop Sci., September 1, 2003; 43(5): 1747 - 1753. [Abstract] [Full Text] [PDF]

				S. Bentolila, A. A. Alfonso, and M. R. Hanson A pentatricopeptide repeat-containing gene restores fertility to cytoplasmic male-sterile plants PNAS, August 6, 2002; 99(16): 10887 - 10892. [Abstract] [Full Text] [PDF]

				J.F. Zhang and J. McD. Stewart Inheritance and Genetic Relationships of the D8 and D2-2 Restorer Genes for Cotton Cytoplasmic Male Sterility Crop Sci., March 1, 2001; 41(2): 289 - 294. [Abstract] [Full Text] [PDF]