Delineation by Fluorescence in Situ Hybridization of a Single Hemizygous Chromosomal Region Associated With Aposporous Embryo Sac Formation in Pennisetum squamulatum and Cenchrus ciliaris

Corresponding author: Peggy Ozias-Akins, P.O. Box 748, University of Georgia, Tifton, GA 31793-0748., ozias{at}tifton.cpes.peachnet.edu (E-mail)

Communicating editor: V. L. CHANDLER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Apomixis is a means of asexual reproduction by which plants produce embryos without meiosis and fertilization; thus the embryo is of clonal, maternal origin. We previously reported molecular markers showing no recombination with the trait for aposporous embryo sac development in Pennisetum squamulatum and Cenchrus ciliaris, and the collective single-dose alleles defined an apospory-specific genomic region (ASGR). Fluorescence in situ hybridization (FISH) was used to confirm that the ASGR is a hemizygous genomic region and to determine its chromosomal position with respect to rDNA loci and centromere repeats. We also documented chromosome transmission from P. squamulatum in several backcrosses (BCs) with P. glaucum using genomic in situ hybridization (GISH). One to three complete P. squamulatum chromosomes were detected in BC₆, but only one of the three hybridized with the ASGR-linked markers. In P. squamulatum and in all BCs examined, the apospory-linked markers were located in the distal region of the short arm of a single chromosome. All alien chromosomes behaved as univalents during meiosis and segregated randomly in BC₃ and later BC generations, but presence of the ASGR-carrier chromosome alone was sufficient to confer apospory. FISH results support our hypotheses that hemizygosity, proximity to centromeric sequences, and chromosome structure may all play a role in low recombination in the ASGR.

APOMIXIS in flowering plants is an asexual mode of reproduction that results in the formation of seeds containing an embryo with the maternal genotype (ASKER and JERLING 1992 ). Apomixis has both sporophytic (adventitious embryony) and gametophytic forms (KOLTUNOW 1993 ). In adventitious embryony a new sporophyte arises from a somatic cell of the ovule. Gametophytic apomixis can result from diplospory, in which the unreduced embryo sac originates from a generative cell, or apospory, in which the embryo sac develops from somatic cells of the ovule (GRIMANELLI et al. 2001 ). After embryo sac development, parthenogenesis occurs in both forms of gametophytic apomixis. Apomixis has been described in at least 33 of the 460 families of angiosperms (CARMAN 1997 ), but has been most frequently observed in the Poaceae, Asteraceae, and Rosaceae (RICHARDS 1986 ).

Apomixis has vast potential for application to breeding and propagation of crop plants, although the potential is far from realized. This reproductive strategy enables the production of clonal seed from a particular genotype, even one that is highly heterozygous, thus immortalizing hybrid vigor. Unfortunately, no cultivated crop displays a sufficient degree of apomixis for practical application, and only a few crop plants have wild relatives that are near-obligate apomicts. Considerable efforts have been made to introduce the trait into crops such as maize (LEBLANC et al. 1995 ; SAVIDAN 2001 ) and pearl millet (DUJARDIN and HANNA 1989 ) by traditional breeding involving interspecific hybridization. The most successful of these hybridization programs has used Pennisetum squamulatum Fresen, an apomictic relative of the domesticated plant, pearl millet [P. glaucum (L.) R. Br.]. Pearl millet is grown for its grain primarily in Africa and India and as a forage crop in tropical and subtropical regions of the world, including the southern United States. The transfer of apomixis from P. squamulatum to P. glaucum has been pursued for the last two decades, and the crossing scheme that led to the recovery of a backcross 3 (BC₃) individual showing near-obligate apomixis has been described (DUJARDIN and HANNA 1989 ). Subsequently, additional backcross generations (BC₄–BC₇) were produced. The transfer of apomixis from P. squamulatum to pearl millet has been hindered by linkage drag of undesired characteristics, among them low seed set (5% for open-pollinated BC₃ and later generations) and high (60%) pollen sterility. Genetic and cytogenetic analyses of these apomictic backcross lines have been carried out to study their chromosome behavior and the inheritance of apomixis (DUJARDIN and HANNA 1989 ; HANNA et al. 1993 ). All of these backcross lines have 27–29 chromosomes, but traditional cytogenetic investigation has not been able to distinguish the alien chromosomes from the pearl millet chromosomes.

An alternative approach to producing apomictic crops might be to transfer to a sexual plant one or more well- characterized genes known to confer the trait of apomixis. In the search for such genes, information about the molecular and genetic basis of apomixis in nondomesticated species is being accumulated. Genetic and molecular mapping studies based on the analysis of offspring from apomictic by sexual crosses have been carried out on multiple species reproducing by gametophytic apomixis (GROSSNIKLAUS et al. 2001 ). Aposporous apomixis in Panicum maximum (SAVIDAN 1982 ), Ranunculus auricomus (NOGLER 1984 ), Cenchrus ciliaris (SHERWOOD et al. 1994 ), Brachiaria decumbens (DO VALLE and MILES 2001 ), Paspalum simplex (CACERES et al. 2001 ), and P. squamulatum (OZIAS-AKINS et al. 1998 ) is inherited as a single, dominant Mendelian factor. Inheritance of diplosporous apomixis in Taraxacum and Erigeron has been shown to be a multilocus phenomenon where the development of diplosporous embryo sacs and parthenogenesis can be genetically separated (VAN DIJK et al. 1999 ; NOYES and RIESEBERG 2000 ).

Our lab has centered its research on two grasses, P. squamulatum Fresen and C. ciliaris L. [syn. P. ciliare (L.) Link; buffelgrass], both of which reproduce by apospory. The phylogeny of the "bristle-grass" clade of panicoid grasses has been investigated recently (DUVALL et al. 2001 ; GIUSSANI et al. 2001 ; DOUST and KELLOGG 2002 ). Current evidence supports the idea that Cenchrus and Pennisetum are paraphyletic genera although both fall in a larger monophyletic clade.

As part of a mapping study, we isolated 12 sequence-characterized amplified region (SCAR) markers that showed no recombination with an apospory-specific genomic region (ASGR) in P. squamulatum (OZIAS-AKINS et al. 1998 ). Ten of the SCARs were conserved and nine showed no recombination with the ASGR in C. ciliaris (ROCHE et al. 1999 ). Several of the SCAR markers were shown by Southern hybridization to be hemizygous with no apparent allele on chromosomal homologs transmitted to the sexual offspring. Bacterial artificial chromosome (BAC) libraries were constructed from the two apomictic lines under study and used for preliminary physical mapping. Classes of BAC clones grouped by apomixis-linked SCAR markers did not overlap, which predicts that building a contig spanning the apomixis locus likely will require multiple walking steps (ROCHE et al. 2002 ). Given the lack of recombination within the ASGR, its apparent large size, and the lack of a high-density genetic map for either species, it became important to visualize the physical location of the ASGR on chromosomal spreads using fluorescence in situ hybridization (FISH) and to trace its transmission to offspring from several backcross generations.

In this article, we present data on the hybridization of metaphase chromosomes with ASGR-linked BAC clones and with pooled ASGR-linked markers to determine the physical location of the ASGR in two apomictic species, P. squamulatum and C. ciliaris, as well as apomictic backcross derivatives from crosses of P. glaucum by P. squamulatum. Furthermore, we compare the location of the ASGR with respect to rDNA loci and centromeric repeats in three genotypes. Finally, we examine the transmission and frequency of the ASGR-carrier chromosome in apomictic backcrosses. FISH results have proved to be instructive in guiding our interpretation of the basis for low recombination, have allowed new insights into the evolution of chromosomal structure in two apomictic grasses, and have determined that a single chromosome is sufficient for the transmission of apomixis and molecular markers linked to the trait.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic stocks:
Pennisetum species used in this study included P. squamulatum (PS26; PI 319196; 2n = 56) and an induced tetraploid pearl millet (P. glaucum, 2n = 4x = 28). Progeny from apomictic and sexual backcross lines used in this study were grown in the field in the summers of 1999 and 2000 (for collecting inflorescences containing various stages of meiosis) and in the greenhouse in the winter of 2000 (for collecting root tips). C. ciliaris, B12-9, is an obligate apomict derived from open pollination of a sexual buffelgrass plant, B2-S (SHERWOOD et al. 1994 ).

Chromosome preparation:
Mitotic chromosome spreads: Root tips were collected and incubated in tap water saturated with -bromonaphthalene for 2–4 hr and subsequently fixed in fresh ethanol:acetic acid (3:1) for a minimum of 2 days before use. Fixed root tips were briefly rinsed in 30 mM citrate buffer, pH 4.5, the root caps were removed, and the meristimatic region was incubated in 0.3% cellulase RS (Karlan Research, Torrance, CA), 0.3% pectolyase Y23 (Karlan Research), and 0.3% cytohelicase (Sigma-Aldrich, St. Louis) in 30 mM citrate buffer, pH 4.5 (ZHONG et al. 1996 ), at 37° for 1.5–3 hr. After a short rinse in the same buffer, the root tips were squashed in 60% acetic acid under a coverslip. The slides were then frozen in liquid nitrogen and the coverslip was removed with a razor blade. The slides were further incubated in 60% acetic acid for 15 min at room temperature, dehydrated in ethanol, and air dried. The slides were stored at -20° until used.

Meiotic chromosome spreads: Inflorescences protruding about one-third of their length from the boot were collected from P. squamulatum, P. glaucum, BC₃, BC₄, BC₅, BC₆, and BC₇. After checking the stage of meiosis in each inflorescence by squashing anthers in acetocarmine, inflorescence sections containing meiotic cells at metaphase I were fixed in ethanol:acetic acid (3:1) and stored at 4°. Up to 1–2 months after fixation, florets were removed from the fixative and soaked in 30 mM citrate buffer (pH 4.5) for 5–10 min. Dissected anthers were cut at the apex and squeezed with a surgical knife to extrude the pollen mother cells (PMCs) into a 10 x 35-mm petri dish containing 1 ml of 30 mM citrate buffer (pH 4.5). PMCs were pipetted into a 1.5-ml microcentrifuge tube where digestion was carried out in 50 µl of enzyme mixture (as above) at 37° for 30–45 min. Digestion time was dependent on the length of time materials had been stored in fixative. Digested PMCs were collected by centrifugation at 600 x g for 5 min at room temperature. The supernatant was removed, and PMCs were resuspended in a volume of 60% acetic acid equal to three times the volume of the digestion solution and incubated for 10 min on ice. PMCs were again collected by centrifugation and resuspended in 4 µl of 60% acetic acid for each slide. Usually one slide was made from each anther by applying the 4 µl of PMC suspension to a precleaned slide and covering it immediately with a 22 x 22-mm cover glass. The cover glass was removed after freezing, and the spread was dehydrated in an ethanol series and air dried.

DNA probes:
Genomic in situ hybridization and pooled marker probes: Genomic DNAs were isolated using the method described in OZIAS-AKINS et al. 1993 and purified with phenol:chloroform:isoamyl alcohol (25:24:1) extraction. Plasmid DNAs containing 12 molecular markers mapped to the ASGR (OZIAS-AKINS et al. 1998 ) were isolated by alkaline lysis minipreps (SAMBROOK et al. 1989 ). Equal amounts (by weight) of DNA from each plasmid were mixed together and 1.5 µg of the mixture was labeled as an ASGR-specific probe.

ASGR-linked probes: The construction of the "polyhaploid" and "buffelgrass" BAC libraries, containing the ASGR from P. squamulatum and C. ciliaris, respectively, has been reported (ROCHE et al. 2002 ). The polyhaploid library was derived from the polyhaploid apomictic line MS228-20 that was germinated from seeds of an open-pollinated apomictic polyhaploid F₁ line derived from a cross between P. glaucum and P. squamulatum (DUJARDIN and HANNA 1986 ). The buffelgrass library was derived from the apomictic line B12-9 (SHERWOOD et al. 1994 ). ASGR-linked BAC clones from both libraries were identified through screening the BAC library filters with ³²P-labeled ASGR-linked SCAR probes. Linkage of individual BACs to the ASGR was confirmed through the analysis of ASGR-specific SCAR or restriction fragment length polymorphism markers (ROCHE et al. 2002 ). To select ASGR-linked BACs containing low-copy sequences for FISH, Southern blots containing 1 µg of HindIII-digested DNA from each BAC clone that had been fractionated on an agarose gel were probed with ³²P-labeled total genomic DNA from BC₃ for the polyhaploid clones or B12-9 for the buffelgrass clones (ZWICK et al. 1997 ). BAC clones that showed little or no signal were used for plasmid preps (QIAGEN, Valencia, CA).

Centromeric probes: Two highly repetitive KpnI fragments of 140 and 160 bp have been reported in the genus Pennisetum (INGHAM et al. 1993 ). The two fragments differ primarily by an 18-bp deletion in the 140-bp KpnI family. A 137-bp HaeIII repeat from P. glaucum cv. Massue has been previously shown to localize to the centromere (KAMM et al. 1994 ). The 160-bp KpnI fragment contains 93% sequence similarity to the 137-bp HaeIII repeat and therefore was expected to hybridize with centromeric sequences. The 160-bp KpnI fragment was used as a centromeric probe for FISH analysis in BC₃ and P. squamulatum. For C. ciliaris, a centromeric probe was obtained by screening the BAC library with a 160-bp KpnI fragment. DNAs from six to eight of the BAC clones that showed the strongest hybridization signals were digested with KpnI and HaeIII. The centromeric sequence content was further confirmed by hybridizing the KpnI- and HaeIII-digested BAC DNAs with a ³²P-labeled 160-bp KpnI clone. The BAC clone containing the largest proportion of laddered fragments was chosen as a centromeric probe for FISH analysis (data not shown).

Ribosomal DNA probe: Plasmid pTA71 containing a 9.5-kb EcoRI 18S-5.8S-25S repeat unit from wheat was used to detect rDNA (GERLACH and BEDBROOK 1979 ).

Probe labeling, slide pretreatment, and hybridization:
DNAs were labeled with either biotin (bio)-11-dUTP (Roche, Indianapolis) or digoxigenin (dig)-11-dUTP (Roche), using a nick translation kit (Roche). Labeled probes were purified by two sequential ethanol precipitations in the presence of 2.1 M ammonium acetate. Pelleted DNA was resuspended in 100% formamide for storage at -20°.

On the day of hybridization, slides were dried at 60° for 30 min. Subsequent pretreatment steps to partially digest RNA and proteins were as described by ZHONG et al. 1996 . Probes were mixed in pairwise combinations for double target hybridization experiments. The hybridization mix for each slide consisted of 1.5–7 ng/µl of probe for each target, 50% formamide, 10% dextran sulfate, 100–300 ng/µl salmon sperm DNA and 2x SSC in a final volume of 10–15 µl. For genomic in situ hybridization (GISH), 100 ng/µl sheared P. glaucum DNA was used as blocking DNA. The hybridization stringency was reduced for the 160-bp KpnI fragment by adjusting the formamide and SSC concentrations to 25% and 5x, respectively. The hybridization mixtures were denatured at 100° for 10 min and snap chilled on ice. The chromosome preparations with the hybridization mix were covered with a coverslip and denatured at 80° for 2 min. Slides were incubated at 60° in a moist chamber for 1 hr and then allowed to cool down to 37° (for FISH) and 39°–42° (for GISH) where they subsequently were incubated for 16–20 hr. Two posthybridization washes were performed in 50% formamide in 2x SSC at 42° for 15 min, except for the 160-bp KpnI fragment where posthybridization washes consisted of 20% formamide in 5x SSC. Formamide washes were followed by room temperature washes in 2x SSC prior to detection.

Probe detection:
Two-color detection was carried out according to ZHONG et al. 1996 with modifications. The dig-labeled probes were detected with FITC by using a fluorescent enhancement kit (Roche) while bio-labeled probes were detected using avidin Texas red (Vector Laboratories, Burlingame, CA). All slides were blocked with 5% nonfat dry milk in 1x PBS (buffer 4M) at 37° for 30 min. All further incubation steps were performed at 37° for 30 min, each followed by three washing steps of 5 min each with T-PBS [0.2% Tween 20 (v/v) in 1x PBS; Fisher Scientific, Pittsburgh] at room temperature. For BAC-FISH, incubation steps were carried out in the following order: Texas red-conjugated avidin; anti-dig IgG from mouse; anti-mouse IgG from sheep conjugated with dig and anti-avidin IgG from goat conjugated with biotin; sheep anti-dig, fluorescein-conjugated fab fragment, and Texas red-conjugated avidin. Since signal amplification was not required for GISH, only the last step was carried out. The slides were given a final rinse in 1x PBS and mounted in Vectashield (Vector Laboratories) containing 4',6-diamidino-2-phenylindole (DAPI; 1.5 ng/µl).

Slides were examined under an Olympus BX50 fluorescence microscope. A minimum of 20 spreads for BAC-FISH and 5 spreads for GISH were examined for each slide. More than 80% of the spreads produced discrete signals. Fluorescent signals were detected for DAPI (_ex = 360 nm, _em = 420 nm), FITC (_ex = 480 nm, _em = 515 nm), and Texas red (_ex = 560 nm, _em = 645 nm), and monochrome digital images were captured with a charge-coupled device camera (SenSys, Photometrics, Tucson, AZ). Images were pseudocolored with blue for DAPI, green for FITC, and red for Texas red. Images were compiled with Image Pro Plus, version 4 for Win 95/98 (Media Cybernetics, Silver Spring, MD). Final adjustments were made using Adobe Photoshop version 5.0. Chromosome lengths were obtained with the image analysis programs, Object-Image2.08 and CHIAS III (KATO and FUKUI 1998 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Generation of stocks segregating for apomixis:
The trait for apomixis was introduced from P. squamulatum into the sexual species P. glaucum according to the recurrent backcrossing scheme diagrammed in Fig 1. Backcross generations 1–7 were generated and screened according to the methods outlined in DUJARDIN and HANNA 1989 . Since a double cross hybrid of (P. glaucum x P. purpureum) x (P. glaucum x P. squamulatum) was used early in the crossing scheme to increase male fertility in the backcrosses (DUJARDIN and HANNA 1984A ), the apomictic double cross also was included in this study. Mode of reproduction could be determined by examining ovules cleared in methyl salicylate for aposporous embyo sac development (YOUNG et al. 1979 ) and by scoring the uniformity of progeny in the field or greenhouse. Since these were heterozygous plants, variable progeny indicated some degree of sexuality.

View larger version (21K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Pedigree summary of hybrids and backcrosses used for FISH/GISH. Materials used for this investigation are shown in boldface type. The F1 of P. glaucum x P. purpureum, BC2, and BC4 were not included in these experiments. Chromosome doubling was accomplished with colchicine treatment.

Comparison of P. squamulatum and P. glaucum chromosomes in interspecific hybrids:
P. squamulatum apparently shares similar DNA sequence with P. glaucum at telomeric and centromeric regions since these regions on the chromosomes of P. squamulatum were blocked with P. glaucum genomic DNA (Fig 2A and Fig B). We observed 56 chromosomes in P. squamulatum, accession PS26 (Fig 2A), contrary to the previous reports of 54 chromosomes in most other accessions (2n = 6x = 54; RAMAN et al. 1959 ; PATIL et al. 1961 ; SISODIA 1970 ; RANGASWAMY 1972 ; DUJARDIN and HANNA 1984B ). SINDHE 1976 observed two supernumerary chromosomes in this species that were rod-like and acrocentric or submetacentric and that were largely eliminated during meiotic cycles. Of the 56 chromosomes that we observed in P. squamulatum accession PS26, none fit the description of the supernumerary chromosomes of SINDHE 1976 . GISH with biotinylated genomic DNA of P. squamulatum confirmed that chromosomes from P. squamulatum were introduced into P. glaucum through crossing of the two distantly related species. Although we examined the chromosome number in only four F₁ individuals, all showed 28 P. squamulatum chromosomes (differentiated by GISH), which suggests that regular disjunction occurred during meiosis and that subsequent transmission to progeny was normal. The origin of the two additional chromosomes in PS26 remains undetermined, but their presence or absence has no apparent effect on apomixis. A total of 42 chromosomes, including 28 P. squamulatum chromosomes and 14 P. glaucum chromosomes, were observed in both apomictic and sexual F₁ individuals (Fig 2B and Fig C, respectively). We used F₁ plants to investigate the chromosome characteristics of P. squamulatum since each F₁ contained half of the chromosomes of P. squamulatum, and chromosomes from both species could be directly compared in F₁ hybrids. The size range in P. squamulatum and P. glaucum was 9.87–3.95 µm and 10.46–5.87 µm, respectively. The length changes in P. squamulatum formed almost a continuous gradient from the largest to the smallest chromosome. Most of the chromosomes were metacentric or submetacentric (Fig 2B and Fig C). It was not possible to unambiguously identify the potential homo(eo)logous chromosomes on the basis of morphology.

View larger version (75K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Results of GISH and FISH on chromosomes from materials in the pedigree shown in Fig 1. White arrowheads indicate chromosomes with satellites. Shown is hybridization of mitotic (A–K) and meiotic (L–O) chromosome preparations with dual-labeled probes of P. squamulatum genomic DNA (red) and the mixture of 12 ASGR-linked markers (green; A–C and E–H); P. squamulatum genomic DNA (red) and P. purpureum genomic DNA (green; D and K); or P. squamulatum genomic DNA alone (green; I, J, and L–O). Because both P. squamulatum and P. purpureum genomic DNAs hybridized to the same chromosomes, the color of the alien chromosomes in D and K appears yellow. Blue chromosomes were blocked with genomic DNA of P. glaucum. (A) P. squamulatum; (B) apomictic F1 290-181D; (C) sexual F1 290-105; (D) BC1; (E) BC3; (F) BC6 43-1; (G) BC5 44-4; (H) BC6 62-1; (I) BC7 60-9; (J) BC5 44-1; (K) BC3; (L) BC3 J34; (M) BC7 J57-3; (N) BC6 J35; (O) BC7 J57-2.

Pearl millet (P. glaucum) is one of the two species in section Penicillaria with a base chromosome number of x = 7 (STAPF and HUBBARD 1934 ; JAUHAR 1981 ). There were seven pairs of chromosomes from P. glaucum in the F₁ hybrid with P. squamulatum. Only three chromosome pairs could be distinguished easily from others using morphological features [1, longest and metacentric; 2, submetacentric; and 7, shortest and subtelocentric with a satellite on the short arm (Fig 2E and Fig K)]. For the P. glaucum chromosomes, the percentage of the total haploid chromosome length in the F₁ ranged from 9.8% (shortest chromosome) to 17.8% (longest chromosome), which is consistent with a previously published karyotype from KHALFALLAH et al. 1993 .

Transfer of P. squamulatum chromosomes into P. glaucum through backcrossing:
Chromosome preparations were made from seven different generations (including P. squamulatum, F₁, BC₁, BC₃, BC₅, BC₆, and BC₇), 20 different lines, and 2–22 plants from each. Both metaphase I of meiosis and mitotic metaphase plates were investigated with the combination of FISH and/or GISH (Table 1). The number of P. squamulatum chromosomes was progressively reduced as backcrossing advanced with the recurrent parent, P. glaucum. In BC₁, differential labeling of genomic DNAs from P. squamulatum and P. purpureum (a bridging species used as a parent prior to the backcrossing step in the introgression program) resulted in hybridization of both probes to the same 14 chromosomes that could be observed by the yellow signal from the combined red and green fluorescence (Fig 2D). This result indicates that there is strong sequence similarity among repeats in the two genomes. Also in BC₁, the number of chromosomes that were blocked with pearl millet DNA increased to 22 (Fig 2D). At the BC₃ generation, three chromosomes and one segment of a fourth chromosome hybridized with P. squamulatum genomic DNA (Fig 2E and Fig L).

Alien chromosome transmission in recurrent backcrosses was determined with a combination of FISH and karyotyping. The characteristics of chromosomes in BC₃ that hybridized with P. squamulatum genomic DNA (alien chromosomes) or were blocked by P. glaucum DNA are shown in Fig 3 and Table 2. Although it is difficult to distinguish between the chromosomes of pearl millet and the alien chromosomes by only the length data (Table 2), the condensation levels revealed upon DAPI staining are substantially different (Fig 3). The ASGR-carrier chromosome has two moderately condensed regions, one around the centromere and another on the short arm. It is possible to recognize this chromosome in BC₃ solely on the basis of its morphology. Genomic DNA of P. purpureum also hybridized to the three to four alien chromosomes of BC₃ when it was used as one of the probes (Fig 2K). Beyond the BC₃ generation, plants classified as apomictic (i.e., producing aposporous embryo sacs, including both obligate and facultative outcomes) contained one to three chromosomes that hybridized with P. squamulatum DNA, and the number of P. glaucum chromosomes ranged from 26 to 28 (Fig 2, F–J). In these materials, the total chromosome number was 28 or 29.

View larger version (38K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Ideogram of BC3 based on five chromosome spreads. Black and gray areas indicate levels of high and moderate condensation, respectively, as determined from DAPI fluorescence. The threshold for high condensation was based on the appearance of the black region on chromosome 7. The threshold for moderate condensation was based on the appearance of the region on the short arm of the ASGR-carrier chromosome.

GISH of meiotic metaphase I indicated that the alien chromosomes in BC₃ and later backcross generations did not pair and thus are presumed not to be homologous to each other. They also were not homologous to any chromosome of P. glaucum because they were exclusively observed as univalents that lagged at anaphase I (Fig 2L and Fig M) or assorted to either pole randomly (Fig 2N and Fig O).

Selection of ASGR-linked BAC clones for FISH:
Twenty-six polyhaploid and 17 buffelgrass ASGR-linked BAC clones that were isolated from the polyhaploid and buffelgrass BAC libraries (ROCHE et al. 2002 ) were tested for their repetitive DNA content by hybridization with total genomic DNA from the respective parental species. Nineteen of the 26 ASGR-linked polyhaploid BAC clones contained a low amount of repetitive DNA. Among the 17 buffelgrass ASGR-linked BAC clones, 12 were low in repetitive DNA content. The BACs used for this FISH study have been summarized in Table 3. The BACs were designated by "p" or "c" if they were derived from the polyhaploid or C. ciliaris, respectively. Fingerprinting analysis showed no overlap in either species between BAC clones containing different SCAR markers (ROCHE et al. 2002 ). In addition, BAC clones carrying the same SCAR designation can be from duplicated regions within the ASGR (ROCHE et al. 2002 ). Each BAC clone for which results are presented produced a discrete signal with a good signal/noise (s/n) ratio without the use of C_ot DNA. Cytoplasm was the major factor affecting the quality of the signal produced. Other factors that affected the s/n ratio included nonspecific binding of antibody and the probe quality.

A single chromosome transmits the ASGR to backcross progeny:
Twelve ASGR-linked molecular markers were pooled, labeled with digoxigenin, and probed simultaneously with labeled P. squamulatum DNA onto the materials as shown in Table 1. Six of the markers were known to be low-copy-number DNAs and the others were repetitive sequences (OZIAS-AKINS et al. 1998 ), but the distribution of the repeats across the genome had not been determined. The FISH signal from these pooled markers was consistently observed at the end of a single, metacentric P. squamulatum chromosome (long arm to short arm ratio of 1.34 ± 0.19; Fig 2A, Fig B, and E–H, green, arrowheads). Plants with the terminal ASGR-specific signal were later classified as either obligate or facultative apomicts. Plants with only meiotically derived embryo sacs did not show the terminal ASGR signal on any chromosome, although other alien chromosomes were sometimes present as shown by GISH. It is likely that the signal from the ASGR-linked markers was derived mainly from the repetitive members of the pool since a probe pool that eliminated the single copy markers gave the same result (data not shown).

Two-color FISH was used to map the comparative positions of ASGR-linked BAC clones containing different SCAR markers. Chromosome spreads from BC₃ were most extensively examined with BAC-FISH. In BC₃, all the ASGR-linked polyhaploid BACs listed in Table 3 localized to the distal end of a single chromosome. FISH results from a subset of the ASGR-linked BAC clones are shown in Fig 4A–C. In each case of dual target mapping, an overlapping (yellow) signal was detected when the FITC (green) and Texas red (red) signals from BAC clones containing different SCAR markers were merged. On highly condensed root-tip metaphase chromosomes of BC₃, the single hybridizing chromosome showed no separation in signal sites between any of the ASGR-linked BAC clones (Fig 4A). Although not directly tested, it is unlikely that the single chromosome that showed signal was derived from pearl millet and not P. squamulatum as the pearl millet background is autotetraploid and therefore should have yielded typically four chromosomes with signal. Because the ASGR in BC₃ was introgressed from P. squamulatum, PS26 (2n = 56), it also was important to test the number and location of hybridization sites in this apomictic parent and to compare the results with BC₃.

View larger version (74K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Localization of ASGR-linked BAC clones on metaphase chromosomes by dual-target mapping with two BACs (A–C), centromeric probes (D–F), and ribosomal DNA (G–I). In A–C, hybridizing sites on the ASGR-carrier chromosome are marked with arrowheads. Images in insets show enlarged ASGR-carrier chromosome. (A) BC3 hybridized with BAC clones p203 (red) and p109 (green); images in inset show signals for p207 (red)/p001 (green) and p303 (green)/p503 (red). (B) P. squamulatum hybridized with p207 (red) and p109 (green). (C) C. ciliaris hybridized with c001 (green) and c101 (red). Overlapping signals in A–C are yellow. (D) BC3 hybridized with p205 (red) and 160-bp KpnI centromere repeat (green). (E) P. squamulatum hybridized with p208 (red) and a 160-bp KpnI centromere repeat (green). (F) C. ciliaris hybridized with c013 (red) and a BAC containing centromere repeats (green). (G) BC3 hybridized with p109 (red) and rDNA (green, arrowheads). (H) P. squamulatum hybridized with p109 (red) and rDNA (green, arrowheads). (I) C. ciliaris hybridized with c101 (red) and rDNA (green, arrowheads). Bars, 10 µm.

To determine if introgression of the ASGR in BC₃ involved any structural rearrangement of the hybridizing P. squamulatum chromosome, FISH with ASGR-linked BAC clones also was performed on P. squamulatum. Only a subset of the ASGR-linked BAC clones [p109 (Q8M) and p207 (ugt197)] were hybridized and detected in pairs using two-color FISH as described above for BC₃. Both BAC clones again hybridized to the distal end of a single chromosome with no detectable signal on other chromosomes (Fig 4B). Similar to BC₃, the two BAC signals overlapped. No detectable structural rearrangement was observed when the hybridizing chromosome was recorded either in its native P. squamulatum genetic background or in that of its introgression host.

In C. ciliaris, apomict B12-9 (2n = 36), the ASGR-linked BAC clones containing Q8M [c001] and ugt197 [c101], produced an overlapping signal on a single chromosome (Fig 4C). Unlike P. squamulatum, the position of the signal on the chromosome arm was not distally located.

Location of the ASGR relative to the centromere:
Many Pennisetum sp. chromosomes are small and highly condensed at mitotic metaphase. Under such circumstances the centromere cannot be visualized reliably by simple DAPI staining. Due to this constraint, centromeric probes were used to locate the centromere. Two-color FISH, with an ASGR-linked BAC clone (biotin; red signal) and centromeric probes (dig; green signal), was carried out in BC₃, P. squamulatum, and C. ciliaris (Fig 4, D–F). Interestingly, two centromeric signals were obtained on the chromosome containing the ASGR in BC₃ (Fig 4D) and P. squamulatum (Fig 4E). One signal was at the primary constriction of the metacentric chromosome while a second signal was seen at the end of the chromosome, distal to the ASGR-linked BAC signal. In C. ciliaris, only a single centromere-related signal, to which the ASGR-linked BAC signal was immediately distal, was observed (Fig 4F).

Location of the ASGR relative to rDNA loci:
Two-color FISH was used to locate ASGR-linked BACs (biotin; red signal) in comparison to rDNA loci (dig; green signal). In BC₃, hybridization of metaphase chromosomes with rDNA produced four strong and four weak signals (Fig 4G). P. squamulatum showed two major and six minor rDNA signals (Fig 4H). The signal for the ASGR-linked BAC containing Q8 [p109] was observed on a different chromosome from those showing the rDNA loci in both BC₃ and P. squamulatum. C. ciliaris showed two major and two minor rDNA signals and, in contrast to BC₃ and P. squamulatum, one of the minor rDNA signals shared the chromosome with the ASGR-linked BAC containing ugt197 (c101). The rDNA signal was located distal to the BAC signal (Fig 4I).

The ASGR has no homologous region in P. squamulatum:
The result of FISH with the ASGR-linked marker mixture and BAC clones as probes against P. squamulatum, the apomictic F₁'s, and backcrosses showed that a single chromosome carried the terminal ASGR-specific signal. No homologous chromosome could be detected with these probes and hybridization conditions (Fig 2 and Fig 4). Thus, among the 56 chromosomes in P. squamulatum (PS26), only 1 chromosome showed the hybridization signal, and F₁ plants with this chromosome were apomictic. No signal from the ASGR-linked markers was observed in any sexual F₁ or backcross plant (Fig 2C; Table 1).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Our previous mapping studies have shown that recombination is repressed in the region of the genome that transmits apomixis (OZIAS-AKINS et al. 1998 ). We also have demonstrated that the ASGR is partially hemizygous in nature (OZIAS-AKINS et al. 1998 ; ROCHE et al. 1999 ); thus, during evolution it has undergone considerable sequence and perhaps structural divergence from orthologous and even allelic regions. In this article we show that transmission of this single distinct chromosome is sufficient to confer apomixis in backcrosses with a sexual species. Additional evidence for hemizygosity and other structural features of the ASGR-carrier chromosome from two apomictic species are reported in this work and their potential to impact recombination is discussed.

One chromosome with the ASGR is sufficient for the expression of apomixis:
In this study, hybridization of multiple ASGR-linked probes to several apomictic genotypes (species, crosses, and backcrosses) detected a single hybridizing chromosome in each of the plants tested. Previous mapping of apomixis-linked molecular markers from an apomictic BC₃ line to a BC₄ population strongly suggested that a single chromosome was necessary and sufficient for expression of the trait (OZIAS-AKINS et al. 1993 ). Similarly, apomixis was mapped as a single-dose allele in P. squamulatum (OZIAS-AKINS et al. 1998 ). The uniqueness of ASGR-linked SCAR sequences (1 kb in size) had been previously shown by Southern blots where hybridization and segregation data led to their description as hemizygous sequences (OZIAS-AKINS et al. 1998 ; ROCHE et al. 1999 ). In this study, FISH probes consisting of entire ASGR-linked BAC clones (80- to 100-kb inserts) also resulted in the observation of hemizygosity where, in all cases, BAC-FISH signal was detected on only a single chromosome in apomictic plants. The ASGR-carrier chromosome, which was ultimately identifiable by morphological characteristics and multiple FISH probes, was always absent from sexual plants even though these plants could carry other chromosomes that hybridized with P. squamulatum DNA by GISH. Although the lack of any detectable signal from ASGR-linked BACs to a homologous chromosome in backcrosses was not unexpected, given that the majority of chromosomes were derived from the sexual recurrent parent, P. glaucum, we anticipated that additional hybridization signals of the ASGR-linked BACs to chromosomes in P. squamulatum or C. ciliaris may occur. Both P. squamulatum and C. ciliaris are polyploid; thus more than two homologous or homeologous chromosomes containing allelic sequences could have hybridized with the BAC clones. In Arabidopsis thaliana, a contig of BAC clones produced allelic hybridization signals on a pair of chromosomes, whereas the same probe produced multiallelic signals in Brassica rapa (JACKSON et al. 2000 ). Similarly, a dihaploid potato clone from tetraploid potato always showed signals on a pair of somatic chromosomes when probed with BAC clones from a potato BAC library (SONG et al. 2000 ).

Our FISH results confirm our earlier genetic and molecular mapping studies, which concluded that apomixis was controlled by a single, dominant "locus" (a locus being defined here as an inherited unit, not necessarily a gene). Since the ASGR does not have a strictly colinear region on the "sexual" homologs in P. squamulatum, it should behave genetically like a dominant, single gene. This inheritance pattern also is consistent with other recent molecular genetic studies, which have contributed to the emergence of a predominant hypothesis that apomixis is regulated by one or two, usually dominant loci (GROSSNIKLAUS et al. 2001 ). Given the absolute correlation between the presence of the ASGR-carrier chromosome and apomictic reproduction, it seems likely that there are apomixis-specific factors in the ASGR, which underlie certain developmental pathways that allow nucellar cells to form aposporous embryo sacs. Alternatively, the ASGR could place a group of reproduction-related genes in a different structural context through an effect on chromatin remodeling, thereby changing gene expression patterns.

While a single chromosome from P. squamulatum is sufficient for the expression of apomixis, we have noted considerable variation in the degree of apomixis (as measured by progeny analysis) among various aposporous backcross lines ranging in chromosome number from 27 to 29 (HANNA et al. 1993 ). These exact materials were not available for FISH analysis, and only small numbers of progeny of one-, two-, and three-chromosome addition/substitution backcross lines from this study were analyzed. Preliminary observations have shown that the BC₇ lines in this study with one, two, or three alien chromosomes can have as few as 40–60% or as many as 85–100% of the pistils producing aposporous embryo sacs. Progeny analysis in these materials generally indicates a correlation between the frequency of embryo sac type and reproductive mode. The facultativeness we have observed also could be affected by environment (W. W. HANNA, personal observations), although there is little published evidence for such effects in Pennisetum (HUSSEY et al. 1991 ). Genetic modifiers of apomixis have been reported to exist in Hieracium (KOLTUNOW et al. 2000 ), and it may eventually be necessary to understand the role of such modifiers in the penetrance of the trait if it is to be utilized in crop plant breeding and genotype maintenance. To carefully analyze the penetrance of the trait in different backcross lines of Pennisetum, a multiyear study that includes both greenhouse and field environments, plus embryo sac clearing and progeny analysis, will need to be conducted.

Structural features of the ASGR-carrier chromosome:
Four major and four minor rDNA signals were observed in BC₃, which has a tetraploid pearl millet background. These results are consistent with earlier reports of two major and two minor rDNA signals for diploid P. glaucum (MARTEL et al. 1996 ; LIU et al. 1997 ). P. squamulatum exhibited two major and six minor rDNA sites, while C. ciliaris contained two major and two minor rDNA sites. The locations of rDNA, with respect to the ASGR, differentiated C. ciliaris from P. squamulatum. Ribosomal DNA sites have been shown to be mobile in species such as wheat and rice (DUBCOVSKY and DVORAK 1995 ; SHISHIDO et al. 2000 ). Translocation of distal regions has been suggested as the origin of the new rDNA loci in some species (ARNHEIM 1983 ; BADAEVA et al. 1996 ). The mechanisms responsible for the apparent mobility of rDNA are not clear and, although transposition or translocation may be involved, it also is possible that cryptic rDNA sites could be amplified by unequal crossing over (DUBCOVSKY and DVORAK 1995 ). Comparing the rDNA location with the ASGR in other apomictic Pennisetum species may provide information about the evolution of the ASGR in these plants.

An abundant 160-bp cloned repeat from P. squamulatum was shown to localize to the centromeres of pearl millet and P. squamulatum as well as to the end of the ASGR-carrier chromosome. The 140-bp repeat from pearl millet shares significant sequence similarity with P. squamulatum repeats (140 and 160 bp), even though there is an additional repeat unit structure in P. squamulatum (INGHAM et al. 1993 ). According to INGHAM et al. 1993 , the smaller repeat family likely diverged from the larger repeat unit by a deletion.

Recombination in the ASGR:
Multiple hypotheses to explain the basis of repressed recombination in the ASGR can be found in the literature (GRIMANELLI et al. 1998 ; OZIAS-AKINS et al. 1998 ; GROSSNIKLAUS et al. 2001 ; ROCHE et al. 2001 ). A prerequisite for recombination is pairing and synapsis of homologous chromosomes. OZIAS-AKINS et al. 1998 and ROCHE et al. 2001 suggested that the apospory locus could be located on a mini- or a B chromosome devoid of any pairing homolog. Although there are reports of B chromosomes in P. squamulatum (SINDHE 1976 ), and indeed the accession we used in this study contained 56 chromosomes, which presumably reflects the presence of two B chromosomes in addition to the normal hexaploid complement of 54 chromosomes, no relationship has been established between the aposporous phenotype and the presence of a B chromosome. Obligate apomixis also has been observed in accessions of P. squamulatum containing only 54 chromosomes (DUJARDIN and HANNA 1984B ). Furthermore, the ASGR-carrier chromosome does not appear to segregate strictly as a univalent (although univalents have been reported in P. squamulatum; SINDHE 1976 ; DUJARDIN and HANNA 1984B ), because amplified fragment length polymorphism markers displaying recombination are linked with, but outside of, the ASGR (Z. B. CHEN, W. W. HANNA and P. OZIAS-AKINS, unpublished data).

Even though the entire ASGR-carrier chromosome does not strictly behave as a B chromosome, the ASGR itself does display characteristics, such as hemizygosity, that set it apart from other regions of the genome. The unusual features of the ASGR could be explained by a combination of male-only transmission following several potential generative processes: (i) introgression of a divergent chromosomal fragment through hybridization such as has been observed through artificial hybridization of sugar beet with wild relatives to transfer nematode resistance (SCHMIDT et al. 1997 ), (ii) the invasion of retrotransposons and their subsequent amplification as in maize knobs (ANANIEV et al. 1998 ), (iii) the de novo assembly of satellite DNA families into a distinct heterochromatic domain with "an ostensibly foreign origin" (LANGDON et al. 2000 ), (iv) the fusion of otherwise free supernumerary chromatin to an autosomal chromosome as previously described in insects (ARAUJO et al. 2001 ), or (v) targeted deletion as part of a selfish maintenance mechanism similar to gametocidal genes (OGIHARA et al. 1994 ). Any of these mechanisms could lead to sequence dissimilarity among homologs and interfere with recombination.

Another hypothesis for repressed recombination has been that the apospory "locus" could be located near the centromere of a chromosome (OZIAS-AKINS et al. 1998 ). Reduced recombination already has been shown around the centromeric regions of various organisms (TANKSLEY et al. 1992 ; DAVIS et al. 1994 ; SHERMAN and STACK 1995 ; ROUND et al. 1997 ; MAHTANI and WILLARD 1998 ) and could be due, at least in part, to the formation of centromeric heterochromatin (KHUSH and RICK 1968 ; RICK 1972 ; COPENHAVER and PREUSS 1999 ). Data from this study support proximity to a centromere as a possible cause of low recombination in the ASGR, at least for C. ciliaris. Even for the ASGR in P. squamulatum, the detection of additional centromeric repeats at the end of the ASGR-carrier chromosome could have consequences on recombination. We do not have any evidence for retention of functionality of the terminal centromeric sequences (such a dicentric chromosome probably would not have survived the backcrossing program with two centromeres of equal strength), although the presence of terminal centromeric sequences alone possibly could alter adjacent chromatin structure and lead to reduced recombination.

A third hypothesis for repressed recombination at the ASGR was the presence of a heterozygous inversion (OZIAS-AKINS et al. 1998 ). Heterozygous inversions are well known to influence apparent recombination rates since a single crossover within an inversion loop formed by the pairing of a paracentric inversion with a normal chromosome would lead to the formation of duplications and deficiencies, including dicentric and acentric chromosomes, neither of which is likely to be transmitted (BURNHAM 1962 ; SCHULZ-SCHAEFFER 1980 ). Such chromosomal aberrations usually result in reduced fertility, although we have observed relatively high pollen viability in locally tested accessions of P. squamulatum (82%; DUJARDIN and HANNA 1984B ) in contrast to 56% sterility observed by SINDHE 1976 . However, paracentric inversions can also cause localized asynapsis (lack of pairing) or desynapsis (precocious separation of the paired chromosomes) without leading to chromosomal aberrations (SCHULZ-SCHAEFFER 1980 ) but still resulting in suppressed recombination at the ASGR. Since the location of the ASGR signal and chromosome morphology were similar in both BC₃ and P. squamulatum, no major rearrangement appears to have survived meiosis during the transfer of the chromosome from P. squamulatum to BC₃. However, a comparison of the position of the ASGR and associated centromeric sequences between P. squamulatum and C. ciliaris does suggest that an inversion probably occurred during evolutionary separation of these two species.

The physical localization data provided by FISH offers explanations for repressed recombination at the ASGR. Repressed recombination makes it possible that apospory is controlled, not by a single gene, but rather by two or more genes that are maintained as an intact genetic unit. Gametophytic apomixis requires at least two universal components: formation of unreduced embryo sacs and the capacity for parthenogenetic development of egg cells. It seems unlikely that a single gene controls both of these characters, although models have been proposed for a single master regulator that could cause the precocious expression of certain genes in apomicts that are normally expressed only after meiosis in sexual plants (PEACOCK 1992 ). The idea of the ASGR representing a haplotype with two or more tightly linked genes becomes more appealing in the wake of recent evidence that diplosporous apomixis is governed by more than one locus, where diplospory and parthenogenesis clearly can be uncoupled in some species (TAS and VAN DIJK 1999 ; VAN DIJK et al. 1999 ; NOYES and RIESEBERG 2000 ). Many examples of multigene or complex loci have been found in both plants and animals. Coadapted gene complexes include examples such as the self-incompatibility loci of Brassica, the t haplotype in mouse, and sex chromosomes. Distinct haplotypes emerge in these cases where genes remain linked in phase by the lack of regional intrachromosomal recombination.

Our goal is the isolation of genes for aposporous apomixis that have been mapped to a single ASGR-carrier chromosome, which is sufficient for the expression of apomixis in the background of tetraploid pearl millet. Although this goal is likely beyond our immediate reach, further elucidation of the finer structure of the ASGR and studies into the possible sequence rearrangements at the locus will give more insights into the evolution of apospory in the Pennisetum/Cenchrus complex. To achieve our long-term goal, fragmentation of the ASGR-carrier chromosome and radiation hybrid mapping may be essential for dissection of the detailed structure of the ASGR to the minimum unit required for aposporous reproduction. Such an experimental approach also should generate materials that could shed light on the relevance of the chromosomal context of this hemizygous region for the expression of apomixis.

	FOOTNOTES

¹ These authors contributed equally to the work.

	ACKNOWLEDGMENTS

We are thankful for the technical training provided by Hans de Jong and Jiming Jiang and for the technical assistance provided by Anne Bell, Evelyn Perry, Jacolyn Merriman, and Freddie Cheek. We also thank Dominique Roche for a critical review of the manuscript. This work was supported by the USDA National Research Initiative Plant Genome Program award no. 99-35300-7691. J.A.C. was partly supported by a DOE-Energy Biosciences research fellowship.

Manuscript received June 22, 2002; Accepted for publication November 19, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ANANIEV, E. V., R. L. PHILLIPS, and H. W. RINES, 1998  Complex structure of knob DNA on maize chromosome 9: retrotransposon invasion into heterochromatin. Genetics 149:2025-2037.[Abstract/Free Full Text]

ARAUJO, S. M. S. R., S. G. POMPOLO, F. PERFECTTI, and J. P. M. CAMACHO, 2001  Integration of a B chromosome into the A genome of a wasp. Proc. R. Soc. Lond. Ser. B 268:1127-1131.[Medline]

ARNHEIM, N., 1983 Concentrated evolution of multigene families, pp. 38–61 in Evolution of Genes and Proteins, edited by M. NEI and R. K. KOEHN. Sinauer Associates, Sunderland, MA.

ASKER, S., and L. JERLING, 1992 Apomixis in Plants. CRC Press, Boca Raton, FL.

BADAEVA, E. D., B. FRIEBE, and B. S. GILL, 1996  Genome differentiation in Aegilops. 1. Distribution of highly repetitive DNA sequences on chromosomes of diploid species. Genome 39:293-306.

BURNHAM, C. R., 1962 Discussions in Cytogenetics. Burgess Publishing, Minneapolis.

CACERES, M. E., F. MATZK, A. BUSTI, F. PUPILLI, and S. ARCIONI, 2001  Apomixis and sexuality in Paspalum simplex: characterization of the mode of reproduction in segregating progenies by different methods. Sex. Plant Reprod. 14:201-206.

CARMAN, J. G., 1997  Asynchronous expression of duplicate genes in angiosperms may cause apomixis, bispory, tetraspory, and polyembryony. Biol. J. Linn. Soc. 61:51-94.

COPENHAVER, G. P. and D. PREUSS, 1999  Centromeres in the genomic era: unraveling paradoxes. Curr. Opin. Plant Biol. 2:104-108.[Medline]

DAVIS, C. R., M. S. KEMPAINEN, and C. R. MCCLUNG, 1994  Correlation of the physical and genetic maps of the centromeric region of the right arm of linkage group III of Neurospora crassa.. Genetics 136:639-641.

DOUST, A. and E. A. KELLOGG, 2002  Inflorescence diversification in the Panicoid "bristle grass" clade (Paniceae, Poaceae): evidence from molecular phylogenies and developmental morphology. Am. J. Bot. 89:1203-1222.[Abstract/Free Full Text]

DO VALLE, C. B., and J. W. MILES, 2001 Breeding of apomictic species, pp. 137–152 in The Flowering of Apomixis: From Mechanisms to Genetic Engineering, edited by Y. SAVIDAN, J. G. CARMAN and T. DRESSELHAUS. CIMMYT, Houston.

DUBCOVSKY, J. and J. DVORAK, 1995  Ribosomal RNA multigene loci: nomads of the Triticeae genomes. Genetics 140:1367-1377.[Abstract]

DUJARDIN, M. and W. W. HANNA, 1984a  Cytogenetics of double cross hybrids between Pennisetum americanum-P. purpureum amphiploids and P. americanum x Pennisetum squamulatum interspecific hybrids. Theor. Appl. Genet. 69:97-100.

DUJARDIN, M. and W. W. HANNA, 1984b  Microsporogenesis, reproductive behavior, and fertility in five Pennisetum species. Theor. Appl. Genet. 67:197-201.

DUJARDIN, M. and W. W. HANNA, 1986  An apomictic polyhaploid obtained from a pearl millet x Pennisetum squamulatum apomictic interspecific hybrid. Theor. Appl. Genet. 72:33-36.

DUJARDIN, M. and W. W. HANNA, 1989  Developing apomictic pearl millet—characterization of a BC₃ plant. J. Genet. Breed. 43:145-151.

DUVALL, M. R., J. D. NOLL, and A. H. MINN, 2001  Phylogenetics of Paniceae (Poaceae). Am. J. Bot. 88:1988-1992.[Abstract/Free Full Text]

GERLACH, W. L. and J. K. BEDBROOK, 1979  Cloning and characterization of ribosomal RNA genes from wheat and barley. Nucleic Acids Res. 7:1869-1885.[Abstract/Free Full Text]

GIUSSANI, L. M., J. H. COTA-SANCHEZ, F. O. ZULOAGA, and E. A. KELLOGG, 2001  A molecular phylogeny of the grass subfamily Panicoideae (Poaceae) shows multiple origins of C-4 photosynthesis. Am. J. Bot. 88:1993-2012.[Abstract/Free Full Text]

GRIMANELLI, D., O. LEBLANC, E. ESPINOSA, E. PEROTTI, and D. GONZALEZ DE LEON et al., 1998  Mapping diplosporus apomixis in tetraploid Tripsacum: One gene or several? Heredity 80:33-39.

GRIMANELLI, D., O. LEBLANC, E. PEROTTI, and U. GROSSNIKLAUS, 2001  Developmental genetics of gametophytic apomixis. Trends Genet. 17:597-604.[Medline]

GROSSNIKLAUS, U., G. A. NOGLER, and P. J. VAN DIJK, 2001  How to avoid sex: the genetic control of gametophytic apomixis. Plant Cell 13:1491-1497.[Free Full Text]

HANNA, W. W., M. DUJARDIN, P. OZIAS-AKINS, E. L. LUBBERS, and L. ARTHUR, 1993  Reproduction, cytology, and fertility of pearl millet x Pennisetum squamulatum BC₄ plants. J. Hered. 84:213-216.[Abstract/Free Full Text]

HUSSEY, M. A., E. C. BASHAW, K. W. HIGNIGHT, and M. L. DAHMER, 1991  Influence of photoperiod on the frequency of sexual embryo sacs in facultative apomictic buffelgrass. Euphytica 54:141-145.

INGHAM, L. D., W. W. HANNA, J. W. BAIER, and L. C. HANNA, 1993  Origin of the main class of repetitive DNA within selected Pennisetum species. Mol. Gen. Genet. 238:350-356.[Medline]

JACKSON, S. A., Z. CHENG, M. L. WANG, H. M. GOODMAN, and J. JIANG, 2000  Comparative fluorescence in situ hybridization mapping of a 431-kb Arabidopsis thaliana bacterial artificial chromosome contig reveals the role of chromosomal duplications in the expansion of the Brassica rapa genome. Genetics 156:833-838.[Abstract/Free Full Text]

JAUHAR, P. P., 1981 Cytogenetics and Breeding of Pearl Millet and Related Species. Alan R. Liss, New York.

KAMM, A., T. SCHMIDT, and J. S. HESLOP-HARRISON, 1994  Molecular and physical organization of highly repetitive, undermethylated DNA from Pennisetum glaucum.. Mol. Gen. Genet. 244:420-425.[Medline]

KATO, S. and K. FUKUI, 1998  Condensation pattern (CP) analysis of plant chromosomes by an improved chromosome image analyzing system. CHIAS III. Chromosome Res. 6:473-479.[Medline]

KHALFALLAH, N., A. SARR, and S. S. YAKOVLEV, 1993  Karyological study of some cultivated and wild stocks of pearl millet from Pennisetum typhoides Stapf et Hubb. and P. violaceum (Lam.) L. Rich. Caryology 46:127-138.

KHUSH, G. and C. RICK, 1968  Cytogenetic analysis of the tomato genome by means of induced deficiencies. Chromosoma 23:452-484.

KOLTUNOW, A. M., 1993  Apomixis: embryo sacs and embryos formed without meiosis or fertilization in ovules. Plant Cell 5:1425-1437.[Free Full Text]

KOLTUNOW, A. M., S. D. JOHNSON, and R. A. BICKNELL, 2000  Apomixis is not developmentally conserved in related, genetically characterized Hieracium plants of varying ploidy. Sex. Plant Reprod. 12:253-266.

LANGDON, T., C. SEAGO, R. N. JONES, H. OUGHAM, and H. THOMAS et al., 2000  De novo evolution of satellite DNA on the rye B chromosome. Genetics 154:869-884.[Abstract/Free Full Text]

LEBLANC, O., D. GRIMANELLI, D. GONZALEZ-DE-LEON, and Y. SAVIDAN, 1995  Detection of the apomictic mode of reproduction in maize-Tripsacum hybrids using maize RFLP markers. Theor. Appl. Genet. 90:1198-1203.

LIU, C. J., I. P. KING, T. S. PITTAWAY, S. ABBO, and S. M. READER et al., 1997  Physical and genetical mapping of rDNA sites in Pennisetum (pearl millet). Heredity 78:529-531.

MAHTANI, M. M. and H. F. WILLARD, 1998  Physical and genetic mapping of the human X chromosome centromere: repression of recombination. Genome Res. 8:100-110.[Abstract/Free Full Text]

MARTEL, E., A. RICROCH, and A. SARR, 1996  Assessment of genome organization among diploid species (2n=2x=14) belonging to primary and tertiary gene pools of pearl millet using fluorescent in situ hybridization with rDNA probes. Genome 39:680-687.

NOGLER, G. A., 1984  Genetics of apospory in apomictic Ranunculus auricomus V. Conclusions. Bot. Helv. 94:411-422.

NOYES, R. D. and L. H. RIESEBERG, 2000  Two independent loci control agamospermy (apomixis) in the triploid flowering plant Erigeron annuus.. Genetics 155:379-390.[Abstract/Free Full Text]

OGIHARA, Y., K. HASEGAWA, and H. TSUJIMOTO, 1994  High-resolution cytological mapping of the long arm of chromosome 5A in common wheat using a series of deletion lines induced by gametocidal genes of Aegilops speltoides.. Mol. Gen. Genet. 244:253-259.[Medline]

OZIAS-AKINS, P., E. L. LUBBERS, W. W. HANNA, and J. W. MCNAY, 1993  Transmission of the apomictic mode of reproduction in Pennisetum: co-inheritance of the trait and molecular markers. Theor. Appl. Genet. 85:632-638.

OZIAS-AKINS, P., D. ROCHE, and W. W. HANNA, 1998  Tight clustering and hemizygosity of apomixis-linked molecular markers in Pennisetum squamulatum implies genetic control of apospory by a divergent locus that may have no allelic form in sexual genotypes. Proc. Natl. Acad. Sci. USA 95:5127-5132.[Abstract/Free Full Text]

PATIL, B. D., M. W. HARDAS, an