A Candidate Recombination Modifier Gene for Zea mays L.

Corresponding author: Claire G. Williams, Texas A&M University, 313 Horticulture Building, College Station, TX 77843-2135., claire-williams{at}tamu.edu (E-mail)

Communicating editor: J. A. BIRCHLER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Maize meiotic mutant desynaptic (dy) was tested as a candidate recombination modifier gene because its effect is manifested in prophase I. Recombination rates for desynaptic (dy) and its wild type were compared in two ways: (1) segregation analysis using six linked molecular markers on chromosome 1L and (2) cytogenetic analysis using fluorescence in situ hybridization (FISH)-aided meiotic configurations observed in metaphase I. Chromosome 1L map lengths among the six linked markers were 45–63 cM for five F₂ dy/dy plants, significantly lower than the wild-type F₂ map distance of 72 cM. Chromosomes 2 and 6 were marked with rDNA FISH probes, and their map lengths were estimated from FISH-adorned meiotic configurations using the expectation-maximization algorithm. Chiasma frequencies for dy/dy plants were significantly reduced for both arms of chromosome 2, for chromosome arm 6L, and for eight unidentified chromosomes. There was a notable exception for the nucleolus-organizing region-bearing arm chromosome arm 6S, where dy increased chiasma frequency. Maize meiotic mutant desynaptic is a recombination modifier gene based on cytogenetic and segregation analyses.

A recombination modification system has taxon-specific characteristics. It is coupled with breeding, mutation rate, and genomic architecture, shaping the dynamics of variability available for selection (see review in OTTO and BARTON 1997 ). The recombination modification system for maize is characterized by high genetic variability in recombination rates (WILLIAMS et al. 1995 ) and extremely uneven recombination per unit of the physical map (CIVARDI et al. 1994 ; DOONER and MARTINIZ-FEREZ 1997 ). Recombination modification mechanisms include supernumerary chromosomes (NEL 1973 ), transposons (BROWN and SUNDERSAN 1991 ), and chromosome rearrangements (CARLSON 1988 ). Isolation of recombination modifier genes (rec genes) is becoming a tractable problem (KOROL et al. 1994 ; TIMMERMANS et al. 1997 ).

Meiotic chromosome pairing and synapsis have been well characterized for maize microsporogenesis (RHOADES 1950 ; CHANG and NEUFFER 1994 ; DAWE et al. 1994 ; BASS et al. 1997 ). Homologous chromosomes pair, recombine during prophase I, and then disjoin to opposite spindle poles at anaphase I. Pairing is initiated immediately before the zygotene without a premeiotic alignment stage (DAWE et al. 1994 ) and coincides with telomeric clustering on the nuclear envelope (BASS et al. 1997 ). Chromosomes are brought together by telomeres, which are bound to the nuclear envelope. Telomeric regions pair first; there is no pairing in the proximal half of the arm at this point (BURNHAM et al. 1972 ). The nucleolus-organizing region (NOR)-bearing short arm of chromosome 6 is spatially separated from other chromosomes, suggesting that the nucleolus is independently positioned by a separate mechanism that overrides the telomere-clustering mechanism (BASS et al. 1997 ). After pairing, recombination is initiated by meiosis-specific double-strand breaks (DSBs, KLECKNER 1996 ). DSB formation and subsequent genetic exchange precede the formation of the synaptonemal complex (HAWLEY and ARBEL 1993 ). Only a few of the DSBs resolve into rec intermediates, and their resolution is guided by an unknown crossover control mechanism. Crossover control is hypothesized to occur via "the imposition and relief of stress" such that formation of additional crossovers is prevented within an affected region (KLECKNER 1996 ).

Chiasmata also serve an important mechanical role in meiosis I. One chiasma per bivalent is required for orderly disjunction. Without at least one chiasma per bivalent, homologues migrate to the same pole and cause inviable aneuploid gametes. The position of the chiasma is also a stabilizing mechanical force. A crossover at a distal chromosomal position is the least stable in securing meiotic disjunction in yeast, humans, and Drosophila (KOEHLER et al. 1996 ; ROSS et al. 1996 ). In either case, nondisjunction in maize microsporogenesis causes aborted pollen grains.

Major modifier genes for recombination are defined as coarse controls (SIMCHEN and STAMBERG 1969 ). The coarse control system has numerous genes that rigidly control the progression of meiotic events. The genetically conserved system is revealed by rare deleterious mutants that are usually recessive in nature (KAUL and MURTHY 1985 ). Meiotic mutations are a source of candidate rec genes, which are part of the coarse control system. Mutations that alter chiasma development and formation are more numerous than other types of meiotic mutants, and they provide a source of variability in the otherwise conserved rec modification system (KAUL and MURTHY 1985 ).

Among known maize meiotic genes (STAIGER and CANDE 1992 ), desynaptic is the most likely recombination modifier. The desynaptic (dy) phenotype is caused by a single recessive gene. The mutant is male specific, semisterile, and easily phenotyped by identifying plants with 75% normal pollen count (NELSON and CLARY 1952 ). The desynaptic produces univalents in male meiocytes, but no viable trisomic progeny from either gamete type have been recovered (NELSON and CLARY 1952 ). One explanation for the occurrence of univalents is the "defective chiasma binding" hypothesis, which postulates that dy undergoes normal crossing over followed by a defect in chiasma binding (MAGUIRE 1978 ). An alternative explanation would be the "defective recombination" hypothesis, where dy reduces recombination to the point where orderly disjunction fails.

The defective-chiasma-binding hypothesis has limited experimental support. A report of normal crossing-over was based on the analysis of a distal knob polymorphism on chromosome 4L (MAGUIRE 1978 ). Although crossing over was normal, dy exhibited a failure of chiasma maintenance in microsporocytes, leading to a high frequency of univalents and rod bivalents at diakinesis. Crossing over and chiasma-binding activity were concluded to have separate genetic controls (MAGUIRE 1978 ). Under the defective-chiasma-binding hypothesis, dy should have normal recombination rates, thus excluding it as a candidate recombination modifier gene. Under the defective-recombination hypothesis, dy should have reduced recombination rates. The latter hypothesis requires introgression of genetic stocks with higher background recombination rates, as well as a test of recombination rates along several chromosomes.

In this study, we developed an F₂ pedigree segregating for dy and tested the effect of dy on recombination. Two methods were used: (1) segregation analysis of linked molecular markers for one linkage group and (2) a novel method of estimating genetic map lengths from in situ meiotic metaphase I configurations.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

F₂ pedigree for segregation analysis:
Two F₂ pedigrees were constructed using a dy/dy paternal parent derived from the same stocks described by MAGUIRE 1978 and MAGUIRE et al. 1991 , which were obtained through the courtesy of the Maize Stock Center. The seed stocks were cytogenetically verified by M. Maguire for the Maize Stock Center and then reverified in our lab. The dy/dy paternal parent (MSC 92-360-7) had 77% normal pollen count, and the maternal parent (NCSU 7299-6) was a multiple-marker inbred line fixed for rare isozyme alleles on chromosome 1L. The maternal parent was developed from local United States and tropical Latin American germplasm with 95–100% normal pollen counts. The two resulting F₁ plants were selfed and produced 120 F₂-1 and 86 F₂-2 plants, respectively.

Mutant classification:
Pollen count was based on 1:1 iodine:glycerol staining. The F₂ plants with 60–80% normal pollen counts were selfed as dy/dy mutants, and F₂ plants with 95–100% normal pollen counts were selfed as wild-type plants (backcrosses failed repeatedly because of asynchronous flowering). Plants with pollen counts outside these ranges were unclassified: 41 of the 46 F₂-1 and 28 of the 36 F₂-2 unclassified plants had pollen counts between 81 and 94%. Classified plants in the F₂-1 and F₂-2 arrays were segregating 8:1 and 3:1 for the wild-type (+/+ or +/dy) and mutant (dy/dy) phenotypes, respectively.

From the F₂-1 array, four dy/dy plants (7480-3, 7478-6, 7474-25, and 7477-12) were also selfed to produce 77–96 F₃ progeny, which were used for segregation analysis of six (isozyme) molecular marker loci linked on chromosome 1L. From the F₂-2 array, one wild type (7469-18 or wt21) and one dy/dy mutant (7466-14 or dy21) produced 73–96 F₃ progeny for segregation analysis. For 7477-12 (dy14), the pollen count and the segregation analysis suggested inconsistent interpretation of its genotype, so selfed progeny from dy14 were grown in a winter greenhouse and verified cytogenetically. Wild-type 7469-18 was also checked with a progeny test in the field.

Cytogenetic analysis was conducted on two plants from the F₂-1 array: a dy/dy mutant (7470-2) and a wild type (7472-18). The same plants could not be used for segregation and cytogenetic analyses because there was insufficient material for meiocyte collection, pollen abortion counts, and pollination on a single plant.

Marker selection and linkage map construction:
Isozyme markers were used because they are expressed early, exhibit codominant inheritance, show conserved locus ordering among maps, and represent gene products at the phenotypic level. Six isozyme loci (Amp1, Mdh4, Pgm1, Adh1, Phi1, and Gdh1) were assayed on all F₂ plants in F₂-1 and F₂-2 arrays in the U.S. Department of Agriculture-Agricultural Research Station laboratory (Raleigh, NC) using etiolated coleoptile tissue electrophoresis as described by STUBER et al. 1988 . All gels were scored by two independent observers. F₂ plants that were polymorphic for all six loci (7.8% of the F₂ plants) were selfed to produce F₃ progeny; F₃ progeny were assayed for the same polymorphic markers.

Linkage maps were constructed using the F₂ intercross option in MAPMAKER 3.0 (LINCOLN et al. 1992 ). Map distances for the F₃ progeny were estimated using the KOSAMBI 1944 map function. Individual and pooled linkage maps were tested for linkage heterogeneity on the basis of MORTON 1956 M-test. The M-test is a maximum likelihood test for linkage heterogeneity for two loci, and it is robust for intermediate recombination values and large family sizes (n > 30, RISCH 1988 ).

Map length estimation using fluorescence in situ hybridization (FISH) and meiotic metaphase I (MI) configurations:
MI configurations of bivalents or multivalents depend on the patterns of the chiasmate (one or more chiasma) and achiasmate (no chiasma) segments. Meiotic MI configuration data provide a basis for estimating the map length for chromosome arms or segments (SYBENGA 1975 ; REYES-VALDES and STELLY 1995 ; REYES-VALDES et al. 1996 ). Map lengths estimated from in situ meiotic MI configurations provide complementary data for segregation analysis (Table 1). The meiotic MI configuration approach provides fast, inexpensive, and comprehensive data relative to segregation analyses, which require rare allele stocks and polymorphic marker loci. Also, the MI configuration approach does not require segregating populations and, thus, has broad appeal to a wider range of eukaryotes.

An arm-specific FISH label can distinguish the two arms of a chromosome so that each arm can be reliably classified as a chiasmate (one or more chiasma) or as an achiasmate (no chiasma). Using a FISH locus that is sufficiently distal from the centromere, four types of disomic MI configurations can be distinguished: (1) ring bivalent, (2) rod bivalent with signals in the middle, (3) rod bivalent with signals at the extremes, and (4) two univalents (Figure 1). The frequencies of these four types of configurations can be collectively analyzed according to the methods outlined by REYES-VALDES et al. 1996 to estimate the frequency at which each arm is chiasmate. The genetic map length for each arm can be estimated by using the CARTER and FALCONER 1951 mapping function, which seems best suited for long chromosome intervals (PASCOE and MORTON 1987 ).

View larger version (14K): In this window In a new window Download PPT slide   Figure 1. Relationship between the chiasma patterns and MI configurations of bivalent pairing with a distal FISH marker. Each line represents two sister chromatids. A and B represent the two arms of a chromosome, and arm B is marked with a noncentromeric FISH marker, which allows distinction of the two arms. X...X between homologous chromosome arms A and B in first column represents a chiasmate condition, i.e., one or more chiasmata. The combinations of the two arms with or without chiasmata can produce four types of disomic MI configurations, i.e., ring bivalent (both arms A and B are chiasmate), rod bivalent with signals at the extremes (A is chiasmate and B is achiasmate), rod bivalent with signals in the middle (A is achiasmate and B is chiasmate), and two univalents (both A and B are achiasmate). The centromeres of the bivalents in the third column are marked with arrows. The top and bottom represent the two opposite poles. In this study, a distal 5S rDNA on maize chromosome 2L and a median 18S-28S rDNA on maize chromosome 6 were used to allow the distinction among the four types of MI configurations of bivalent pairing of chromosomes 2 and 6, respectively. The frequencies of these various types of MI configurations were used to estimate the average chiasmate arm frequencies and map lengths for chromosomes 2 and 6 according to the methods outlined by REYES-VALDES et al. 1996 .

We used FISH probes for chromosome-specific analysis of chromosomes 2 and 6. Maize chromosome 2 has a distal 5S rDNA tandem repeat on its long arm, and chromosome 6 has an 18S–28S rDNA tandem repeat at the NOR near the middle of the short arm (COE 1996 ). Therefore, the distal 5S rDNA FISH marker on chromosome 2 and the medial 18S-28S rDNA FISH marker on chromosome 6 were used to allow recognition of the four different disomic MI configurations of chromosomes 2 and 6, respectively (Figure 1). Dual-color FISH with rDNAs as probes allowed the recognition of chromosomes 2 and 6 at the same time.

Chromosome preparation and FISH:
Immature tassels were fixed in 3:1 (v/v) ethanol:acetic acid, hydrated, and squashed in acetocarmine at ca. 80°. Slides were frozen in liquid nitrogen and stored in a freezer at -135°. Probe labeling, dual-color FISH, and photography followed the procedures described in JI et al. 1997 . Plasmid pGmr3, containing the EcoRI fragment of the 18S-28S ribosomal DNA repeat of Glycine max (kindly supplied by E. Zimmer), and plasmid pAm033, containing the 470-bp BamHI fragment of the 5S ribosomal DNA repeat of Acacia melanoxylon (kindly supplied by R. Appels) were labeled with biotin-14-dATP (BioNick kit; Bethesda Research Laboratories, Gaithersburg, MD) and digoxigenin-11-dUTP (Boehringer Mannheim, Indianapolis, IN), respectively. Slides were screened with an Olympus AX-70 microscope equipped with 4',6-diamidino-2-phenylindole (DAPI), fluorescein isothiocyanate (FITC)-DAPI, and DAPI-FITC-rhodamine filter sets.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Maize meiotic mutant dy reduced recombination on multiple chromosome arms. The defective recombination hypothesis was supported by both segregation and cytogenetic analyses. The notable exception was increased recombination along the NOR-bearing region of chromosome 6S. At least one mutant had reduced recombination without nondisjunction and pollen abortion.

Segregation analysis on chromosome 1L:
The segregation data for chromosome 1L revealed reduced map distances for all five dy/dy plants (Figure 2). The cumulative effect for the chromosome arm was stable. This contrasted with the variability among plants and among the five marker intervals (Figure 2). Map distances in the dy plants within the F₂-1 and F₂-2 arrays ranged from 45 to 63 cM, substantially lower than in the wild-type F₂-2, which have distances of 72 ± 2–4 cM (Figure 2). The linkage heterogeneity test was statistically significant at P 0.005 level (² = 40.9 for d.f. = 5) for pooled wild-type and dy maps, indicating linkage heterogeneity among mutant and wild-type maps. No transmission ratio distortion or trisomic loci were observed at any marker loci in F₂ plants or F₂ progeny.

View larger version (21K): In this window In a new window Download PPT slide   Figure 2. Map distances (in centimorgans) among the six isozyme molecular markers derived from the segregation linkage analysis (wt, wild type; PC, pollen count, i.e., the percentage of normal pollens). Gene order is as follows: Amp1, Mdh4, Pgm1, Adh1, Phi1, and Gdh1.

Mutant plant dy14 was originally classified as wild-type by pollen count (95–100% normal pollen). However, segregation analysis revealed reduced recombination (45.7 cM) for dy14 (Figure 2). Classification via a progeny test showed dy14 to be dy/dy (Table 2). If dy14 could be misclassified using pollen abortion counts, then the mutant reduced crossing over without producing univalents, nondisjunction, and pollen abortion. Given the underrepresentation of other dy/dy phenotypes in the F₂-1 array only (segregation ratio of 8:1 for wild-type to mutant, ² = 6.42 for d.f. = 1), then other plants in this progeny array may also be misclassified dy/dy homozygotes. Environmental effects seem unlikely, given that the F₂-2 array in the same planting had the expected 3:1 segregation ratio for the wild-type and mutant plants. Wild-type 7469-18 (wt21) was also progeny tested, but the segregation pattern revealed wt21 to be a wild-type heterozygote (Table 2).

Genome-wide cytogenetic analysis:
Meiocytes having well-spread chromosomes and good chromosome morphology were scored for the occurrence of ring bivalents, rod bivalents, and univalents for all chromosomes (Figure 3A and Figure B). The cytogenetic analysis showed reduced recombination throughout the genome, with the notable exception of the NOR-bearing arm of chromosome 6. The F₂-1 dy/dy progeny had reduced crossing over and more univalents than the F₂-1 wild-type progeny. The average pairing behavior of a meiotic cell and the average achiasmate arm frequency were calculated for the dy mutant and the wild type, first for the whole genome (Table 3) and then for all chromosomes except chromosomes 2 and 6 (Table 3). In both cases, the average achiasmate arm frequencies for the dy mutant were significantly higher than those for the wild type. The average arm achiasmate frequency in the dy mutant was significantly lower for chromosomes 2 and 6 than for the average of the other eight unidentified chromosomes.

View larger version (39K): In this window In a new window Download PPT slide   Figure 3. Photomicrographs of maize meiotic MI chromosome spreads after FISH with 5S and 18S-28S rDNA probes. (A) DAPI image of a microsporocyte at MI from a dy mutant showing five ring bivalents (arrows) and five rod bivalents (arrowheads). (B) Dual-color FISH image of a microsporocyte at MI from maize wild type showing two rod bivalents, with chromosome 6 carrying the green FITC signals (biotinylated 18S-28S rDNA, arrow) and eight ring bivalents, with chromosome 2 carrying the red rhodamine signals (digoxigenin-labeled 5S rDNA, arrowhead). (C–F) Photomicrographs of maize chromosome 6 disomic MI configurations showing formation of chiasmata in different regions. (C) Chromosome 6 MI ring bivalent of a dy mutant with a pair of digoxigenin-labeled 18S–28S rDNA FISH signals (arrows). This bivalent corresponds to the configuration (the third column in Figure 1) with chiasmata in arms A and B. (D) Chromosome 6 MI rod bivalent of dy mutant with a pair of digoxigenin-labeled 18S-28S rDNA FISH signals at the extremes (arrows). This bivalent corresponds to the configuration (the third column in Figure 1), with chiasma(ta) only in arm A. (E) Chromosome 6 MI rod bivalent of dy mutant with a pair of FITC-labeled 18S–28S rDNA FISH signals at the middle (arrows). This bivalent corresponds to the configuration (the third column in Figure 1), with chiasma(ta) only in arm C. (F) Dual-color FISH to MI of maize dy mutant showing that NOR-bearing chromosome 6 forms univalents (marked with FITC-labeled 18S-28S rDNA, arrows).

A closer look at the cytogenetic analysis supports a consistent decrease in crossing over, even if the univalents are excluded from the analysis. At the whole-genome level, the mutant had 5439 chiasmata, representing an average of 17.43 chiasmata per cell or 1.743 cells per bivalent (Table 3A). The wild-type had 3105 chiasmata, 18.16 chiasmata per cell or 1.816 per bivalent (Table 3A). Thus, the dy mutant had 0.073 chiasmata per bivalent less than the wild type. The genome-wide difference becomes slightly greater (from 0.07 to 0.11) if chromosomes 2 and 6 are removed (Table 3B). In Table 3B, the mutant brought about major shifts in the meiocyte class freqencies. The modal class (eight ring + two rod bivalents) decreased 26%, and the next two classes (seven ring + three rod bivalents, six ring + four rod bivalents) jumped 98 and 229%, respectively. On a minor point, cells with univalents tended to have achiasmate arms, suggesting a cell-specific response to dy's defect.

Chromosome 2 MI configurations:
The distal 5S rDNA FISH marker on chromosome 2L allowed the distinction of four different MI configurations of chromosome 2 bivalent pairing (Figure 1; data not shown). The expected frequency for rod bivalents and univalents for the wild type was zero in the given sample (n = 380 cells, Table 4), indicating that achiasmate arm frequency for chromosome 2 was near zero. Compared to the average achiasmate arm frequency of 9.2% for the whole-maize genome (Table 3), the low achiasmate arm frequency for chromosome 2 indicates a slight difference from the genome-wide average, even if univalents are excluded from the analysis.

Chiasmate arm frequencies for maize chromosome 2 were estimated from the frequency of various MI configurations for the dy mutant and wild type (Table 4) according to the method of REYES-VALDES et al. 1996 . Chiasma frequencies for arms A and B of chromosome 2 were both 1 (95% confidence interval of 0.989–1.0) in the wild type (Table 5). Chiasma frequencies for arms A and B of the dy mutant were 0.96 (95% confidence interval of 0.935–0.977) and 0.97 (95% confidence interval of 0.945–0.985), respectively, which were significantly lower than those for the wild type (Table 5). The Carter-Falconer map lengths were 67.7 and 71.8 cM for chromosome 2 arms A and B in dy, respectively, and >88.5 cM for both arms in the wild type (Table 5). The total map lengths of chromosome 2 were 139.5 and >177 cM for the dy mutant and wild type, respectively (Table 5). There was a reasonable agreement between our estimate of >177 cM for the wild type and the estimate of 181.5 cM in the consensus map for chromosome 2 (DAVIS et al. 1996 ).

Chromosome 6 MI configurations:
The median 18S-28S rDNA FISH marker on chromosome 6S allows distinction of four different MI configurations of chromosome 6 bivalent pairing (Figure 1 and Figure 3, C–F). Chiasmate arm frequencies for maize chromosome 6 were estimated from the frequencies of various MI configurations for the dy mutant and wild type (Table 4) according to the method of REYES-VALDES et al. 1996 . The estimated chiasma frequency was lower for dy than for the wild type: 0.98 vs. 1.00 in chromosome 6L (Table 5).

The notable exception was the chiasma frequency estimated for chromosome 6S in dy (0.46), which was significantly higher than in the wild type (0.12, Table 5). The Carter-Falconer map distance estimates for dy were 76.2 cM, 23.2 cM for chromosome 6 arms A and B, respectively, and >89.2 cM, 6.1 cM in the wild type (Table 5). The total map lengths for chromosome 6 were 99.4 and >95.3 cM for the dy mutant and wild type, respectively (Table 5).

Chromosome 6 of the wild type was estimated to be 89.2 cM for its long arm and 6.1 cM for its short arm, which gave a total length of 95.3 cM. This result is shorter than the 144.2 cM length of the most recently published map of chromosome 6 (DAVIS et al. 1996 ). Two reasons may account for this difference. First, we can only estimate its lower limit of chromosome length, which was 89.2 cM because the observed chiasma frequency of the long arm was 1.00 in the sample. The actual arm length may be much longer than its lower limit, thus extending the map length. Second, chromosome arm 6L is long and may have two or even more chiasmata per meiosis. Our estimation was based on the frequency of chiasmate arms and, thus, may underestimate the chiasma frequency. There was reasonable agreement between previous reports for chromosome 6S map length of 3–8 cM (HELENTJARIS et al. 1995 ) and our estimate of 6.1 cM.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The mutation desynaptic reduces recombination on several chromosome arms. Reduced recombination is supported by both segregation and cytogenetic analyses. The notable exception to dy's widespread effect is an increase in recombination in the NOR-bearing chromosome 6S arm. Reduced recombination in desynaptic can occur without nondisjunction and pollen abortion. A crossover control model is proposed. There is strong support for desynaptic as a recombination modifier gene.

Reconciling the defective-chiasma-binding hypothesis:
The defective-chiasma-binding hypothesis states that the mutant has normal crossing over, but that univalents result from a chiasma-binding defect. Supporting evidence for this hypothesis is based on a distal heterochromatic knob polymorphism on chromosome arm 4L (MAGUIRE 1978 ). In this previous study, the dy mutant had the same proportion of microsporocytes with chromosome 4L in crossover configuration as the wild type, yet the proportion of univalents was 20% (MAGUIRE 1978 ). The mutant's effect on recombination is apparent without considering univalent formation (Table 3A and Table 3B).

Our results do not support the defective chiasma binding hypothesis, and the following explanation for MAGUIRE 1978 is offered. The use of the heterochromatic knob as a marker could upwardly bias crossover frequencies compared to the rest of the genome. Use of a genetic stock with low recombination rates (which would increase the proportion of univalents) might lead one to conclude that dy crossover rates were normal and that chiasma binding was defective. By contrast, our data are based on nonknob cytological and molecular markers, and our dy stock was crossed into a high-recombination background, the multiple-allele stock 7299-6, which has a tropical origin. This maternal parent had a higher recombination rate, based on a previous survey of recombination variability (WILLIAMS et al. 1995 ) and on the cytological observation of ring bivalents in the segregating F₂-1 and F₂-2 arrays. Ring bivalents, signifying two or more crossovers, were prevalent in our study, yet they were notably absent in previous dy studies (NELSON and CLARY 1952 ; MAGUIRE 1978 ).

Heterochromatic knobs could have upwardly biased crossing over rates in the previous study (MAGUIRE 1978 ). Knobs are large blocks of constitutively condensed heterochromatin found at specific sites throughout the maize genome. Knobs, regardless of position, directly alter recombination rates. Recombination modification is more pronounced in heterozygous knobs (KIKUDOME 1959 ; CHANG and KIKUDOME 1974 ). Using a knob as a cytological marker, one might erroneously conclude that dy has normal crossing over.

Second, dy may have been initially bred into a genetic stock with low recombination rates, thereby increasing the proportion of univalents and causing chiasma binding to appear consistently defective. MAGUIRE et al. 1991 reported using a genetic stock that had 4% univalents in the wild-type plants and 20% univalents in the mutant, compared to our estimates of 10% univalents in dy and no univalents for the wild-type plants. If so, the original dy genetic stock may have had the minimum number of crossovers for orderly disjunction. Reduced recombination in dy would have lowered crossovers below the minimum required for orderly disjunction, causing a high proportion of univalents.

In our segregating dy populations, mutant and wild-type F₂ plants have a higher proportion of ring bivalents than the populations studied by NELSON and CLARY 1952 and MAGUIRE 1978 . This may be attributed to introgression of tropical maize sources into the segregating dy pedigree; recombination rates on chromosome 1L were substantially higher for tropical backcrosses (WILLIAMS et al. 1995 ). If a genetic background has an excess of crossovers per bivalent over the minimum required for orderly disjunction, then a quantitative reduction in crossovers in a dy/dy plant would be less likely to result in univalent formation, nondisjunction, and pollen abortion.

The desynaptic mutant can reduce recombination without causing nondisjunction, univalents, and pollen abortion:
Our data support the assertion that genetic control of crossing over and chiasma maintenance are independent (MAGUIRE 1978 ; CHUA and ROEDER 1997 ; CONRAD et al. 1997 ). The mutant dy can reduce recombination without a concomitant defect in chiasma binding. The F₂-1 plant (dy14) was originally classified as a wild-type plant in the absence of pollen abortion. The plant showed very low recombination along chromosome 1L (45.7 cM) despite normal pollen counts of 95–100%. A progeny test of dy14's selfed progeny showed dy14 to be a mistyped dy mutant, and it indicated a crude relationship between pollen abortion and univalent formation (Table 2). In this case, desynaptic reduced crossing over without producing univalents, nondisjunction, and pollen abortion. Reduced recombination appears to be the consistent and primary effect. The mutant's cumulative effects later in meiosis are inferred to be indirect, uncoupled from reduced recombination.

In genetic backgrounds with high chiasma frequency, dy can reduce recombination without causing nondisjunction, univalents, and pollen abortion. By contrast, dy in a genetic background with low chiasma frequency would exhibit few crossovers, often dropping below the minimum requirement and producing univalents with a higher frequency. Maize mutant desynaptic is not likely to be a homologue to the tam1 meiotic mutant in yeast (CHUA and ROEDER 1997 ; CONRAD et al. 1997 ) because tam1 exhibits nondisjunction after normal recombination.

The mutant desynaptic increased recombination in the NOR region of chromosome 6S:
Chiasma frequencies were also significantly reduced in both arms of chromosome 2, which carries the 5S rDNA site, and in the non-NOR-bearing (long) arm of chromosome 6, while that of NOR-bearing (short) arm of chromosome 6 was increased significantly in the dy mutant. In the wild type, the chiasmate frequency of the NOR-bearing arm (6S) was extremely low (0.12). The NOR-bearing arm of chromosome 6 is not only physically short, but it also tends to have pairing difficulties around the NOR in the wild type, thus leading to the typical "recombination shadow" (SYBENGA 1975 ). In the dy mutant, the chiasmate frequency of the NOR-bearing arm increased significantly (from 0.12 to 0.46), while the opposite non-NOR-bearing arm of chromosome 6 had a significantly reduced chiasmate frequency (from 1 to 0.98).

The observed chiasmate arm frequencies for chromosome 6 suggest three models. First, recombination for the NOR vicinity may occur later than for the rest of the maize chromosomes. Telomere clustering is associated with active cytoplasmic motility forces (BASS et al. 1997 ), with possible independent locomotion for the NOR-bearing region (HIRAOKA 1952 ). Repressed crossovers in other chromosomes might result in a subsequent compensatory increase for chromosome 6S. Second, dy may alleviate crossover suppression near the NOR region, thus leading to normal crossing over. If so, other repressed regions of the chromosomes, such as telomeres and centromeres, may also exhibit increased recombination. Third, dy may exert a general compensatory effect across all chromosomes, which predicts that other untested chromosome arms have increased recombination. Of the three models, the last seems the least likely. Four out of five chromosome arms had reduced recombination, and the average genome-wide estimate of map length also decreased.

A proposed crossover control model for desynaptic:
Maize meiotic mutant desynaptic is proposed as a crossover control defect. Pairing is normal in dy cells (NELSON and CLARY 1952 ). Recent results indicate that dy is not a telomeric clustering or bouquet formation mutant (H. W. BASS, personal communication). The defective product seems to exert an effect after telomeric clustering and pairing, but before synaptonemal complex formation. The desynaptic mutant has a full synaptonemal complex (MAGUIRE et al. 1991 ). There is some slight reduction in the frequency of synaptonemal complex twists (MAGUIRE et al. 1991 ), and this may be attributed to either a submicroscopic dy defect or to spread preparation protocols.

We propose that desynaptic is a point defect in crossover control, perhaps a defective attachment protein that is critical to a DSB or a defect in crossover control. Crossover control is the process by which DSBs resolve into rec intermediates before or within the Holliday junction. Crossover control occurs after pairing, but possibly before synaptonemal complex formation (STORLAZZI et al. 1996 ).

Our preliminary mapping of dy's location has ruled out chromosome 1L and chromosome 8 (C. G. WILLIAMS, unpublished data), confirming that dy is not the male-sterile mutant ms8 (ALBERTSON and PHILLIPS 1981 ). The desynaptic mutant is not allelic to asynaptic (NELSON and CLARY 1952 ). Cytological and fertility analysis indicates that desynaptic has a male-specific phenotype (NELSON and CLARY 1952 ; CURTIS and DOYLE 1992 ), but it is possible that dy reduces recombination in megasporogenesis. No univalents or trisomic gametes have been observed in megaspore mother cells (CURTIS and DOYLE 1992 ; NELSON and CLARY 1952 ), but cytological and segregation analyses using female dy plants are needed to test for reduced recombination.

In summary, desynaptic appears to be a coarse control mutant with a genome-wide reduction in recombination. The notable exception is an increase in recombination in the NOR-bearing region of chromosome 6S, a region that typically does not recombine. We propose that dy is a crossover control defect with an indirect and sporadic effect on chiasma binding, univalent formation, and nondisjunction.

	ACKNOWLEDGMENTS

The authors gratefully acknowledge the Texas Agricultural Experiment Station, Texas Higher Education Coordinating Board, the McKnight Foundation, and Dr. Charles Stuber (U.S. Department of Agriculture-Agricultural Research Station, Raleigh, NC) for generously facilitating isozyme data collection. This research was sponsored by a Tom Slick Senior Graduate Fellowship at Texas A&M University awarded to Y.J. The programming language used for calculations was MATHEMATICA (version 2.2.3) from Wolfram Research, Inc., using the program CHIASMA.

Manuscript received August 5, 1998; Accepted for publication October 5, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALBERTSON, M. C. and R. L. PHILLIPS, 1981  Developmental cytology of 13 genetic male-sterile loci in maize. Can. J. Genet. Cytol. 23:195-208.

BASS, H. W., W. F. MARSHALL, J. W. SEDAT, D. A. AGARD, and W. Z. CANDE, 1997  Telomeres cluster de novo before the initiation of synapsis: a three-dimensional spatial analysis of telomere positions before and after meiotic prophase. J. Cell. Biol. 137:5-18[Abstract/Free Full Text].

BROWN, J. and V. SUNDERSAN, 1991  A recombination hotspot in the maize A1 intragenic region. Theor. Appl. Genet. 81:185-188.

BURNHAM, C. R., J. T. STOUT, W. H. WEINHEIMER, R. V. KNOWLES, and R. L. PHILLIPS, 1972  Chromosome pairing in maize. Genetics 71:111-126[Abstract/Free Full Text].

CARLSON, W., 1988 The cytogenetics of corn, pp. 259–331 in Corn and Corn Improvement, Vol. 18, Ed. 3, edited by G. S. SPRAGUE and J. DUDLEY. Crop Science Society, Madison, WI.

CARTER, T. C. and D. S. FALCONER, 1951  Stocks for detecting linkage in the mouse, and the theory of their design. J. Genet. 50:307-323.

CHANG, C. C. and G. Y. KIKUDOME, 1974  The interaction of knobs and B chromosomes of maize in determining the level of recombination. Genetics 77:45-54[Abstract/Free Full Text].

CHANG, M. T., and M. G. NEUFFER, 1994 Chromosomal behavior during microsporogenesis, pp. 460–475 in The Maize Handbook, edited by M. FREELING and V. WALBOT. Springer-Verlag, New York.

CHUA, P. R. and G. S. ROEDER, 1997  Tam1, telomere-associated meiotic protein, function in chromosome synapsis and crossover interference. Genes Dev. 11:1786-1800[Abstract/Free Full Text].

CIVARDI, L., Y. XIA, K. EDWARDS, P. S. SCHNABLE, and B. J. NIKOLAU, 1994  The relationship between genetic and physical distances in the cloned a1-sh2 interval of the Zea mays genome. Proc. Natl. Acad. Sci. USA 91:8268-8272[Abstract/Free Full Text].

COE, E., 1996  Gene list and working maps. Maize Genet. Coop. Newslett. 67:133-166.

CONRAD, M. N., A. M. DOMINGUEZ, and M. E. DRESSER, 1997  Ndj1p, a meiotic telomere protein required for normal chromosome synapsis and segregation in yeast. Science 276:1252-1255[Abstract/Free Full Text].

CURTIS, C. A. and G. DOYLE, 1992  Production of aneuploid and diploid eggs by meiotic mutants of maize. J. Heredity 83:335-341[Abstract/Free Full Text].

DAVIS, G., M. MCMULLEN, E. COE, and M. POLACCO, 1996  Working maps. Maize Genet. Coop. Newslett. 70:118-132.

DAWE, R. K., J. W. SEDAT, D. A. AGARD, and W. Z. CANDE, 1994  Meiotic chromosome pairing in maize is associated with a novel chromatin organization. Cell 76:901-912[Medline].

DOONER, H. K. and I. M. MARTINIZ-FEREZ, 1997  Recombination occurs uniformly within the bronze gene, a meiotic hotspot in the maize genome. Plant Cell 9:1633-1646[Abstract].

HAWLEY, R. S. and T. ARBEL, 1993  Yeast genetics and the fall of the classical view of meiosis. Cell 72:301-303[Medline].

HIRAOKA, T., 1952  Observational and experimental studies of the meiosis in special reference to the bouquet stage. XI. Locomotory movement of the nucleolus in the bouquet stage. Cytologia 17:201-209.

HELENTJARIS, T., C. BAYSDORFER, R. FERL, and E. ALMIRA, 1995  Clone identification by sequence similarity and mapping results. Maize Genet. Coop. Newslett. 69:232-256.

JI, Y.-F, D. A. RASKA, T. D. MCKNIGHT, N. M. ISLAM-FARIDI, and C. F. CRANE et al., 1997  Use of meiotic FISH for identification of a new monosome in Gossypium hirsutum L. Genome 40:34-40[Medline].

KAUL, M. L. H. and T. G. K. MURTHY, 1985  Mutant genes affecting higher plant meiosis. Theor. Appl. Genet. 70:449-466.

KIKUDOME, G. Y., 1959  Studies on the phenomenon of preferential segregation in maize. Genetics 44:815-831[Free Full Text].

KLECKNER, N., 1996  Meiosis: how could it work? Proc. Natl. Acad. Sci. USA 93:8167-8174[Abstract/Free Full Text].

KOEHLER, K. E., R. S. HAWLEY, S. SHERMAN, and T. HASSOLD, 1996  Recombination and nondisjunction in humans and flies. Hum. Mol. Genet. 5:1495-1504[Abstract].

KOROL, A. B., I. A. PREYGEL and S. I. PREYGEL, 1994 Recombination Variability and Evolution. Chapman & Hall, London.

KOSAMBI, D. D, 1944  The estimation of map distances from recombination values. Ann. Eugen. 12:172-175.

LINCOLN, S. E., M. DALY and E. LANDER, 1992 Constructing genetic maps with MAPMAKER/EXP 3.0, Ed. 3, pp. 1–73. Whitehead Institute Technical Report, Cambridge, MA.

MAGUIRE, M. P., 1978  Evidence for separate genetic control of crossing over and chiasma maintenance in maize. Chromosoma 65:173-183.

MAGUIRE, M. P., A. M. PAREDES, and R. W. RIESS, 1991  The desynaptic mutant of maize as a combined defect of synaptonemal complex and chiasma maintenance. Genome 34:879-887[Medline].

MORTON, N. E., 1956  The detection and estimation of linkage between genes for elliptocytosis and the Rh blood type. Am. J. Hum. Genet. 8:80-96[Medline].

NEL, P. M., 1973  Modification of crossing over in maize by extraneous chromosomal elements. Theor. Appl. Genet. 43:196-202.

NELSON, O. E. and G. B. CLARY, 1952  Genetic control of semisterility in maize. J. Heredity 43:205-210[Free Full Text].

OTTO, S. P. and N. H. BARTON, 1997  The evolution of recombination: removing the limits to natural selection. Genetics 147:879-906[Abstract].

PASCOE, L. and N. E. MORTON, 1987  The use of map functions in multipoint mapping. Am. J. Hum. Genet. 40:174-183[Medline].

REYES-VALDÉS, M. H. and D. M. STELLY, 1995  A maximum likelihood algorithm for genome mapping of cytogenetic loci from meiotic configuration data. Proc. Natl. Acad. Sci. USA 92:9824-9828[Abstract/Free Full Text].

REYES-VALDÉS, M. H., Y.-F. JI, C. F. CRANE, J. F. TAYLOR, and M. N. ISALM-FARIDI et al., 1996  ISH-facilitated analysis of meiotic bivalent pairing. Genome 39:784-792.

RHOADES, M., 1950  Meiosis in maize. J. Heredity 41:58-67.

RISCH, N., 1988  A new statistical test for linkage heterogeneity. Am. J. Hum. Genet. 42:353-364[Medline].

ROSS, L. O., R. MAXFIELD, and D. DAWSON, 1996  Exchanges are not equally able to enhance meiotic chromosome segregation in yeast. Proc. Natl. Acad. Sci. USA 93:4979-4983[Abstract/Free Full Text].

SIMCHEN, G. and J. STAMBERG, 1969  Fine and coarse controls of genetic recombination. Nature 222:329-332[Medline].

STAIGER, C. J., and W. Z. CANDE, 1992 Cytoskeletal analysis of maize meiotic mutants, pp. 157–171 in Molecular and Cell Biology of the Plant Cell Cycle, edited by J. C. ORMROD and D. FRANCIS. Kluwer Academic Publishers, Dordrecht, The Netherlands.

STORLAZZI, A., L. XU, A. SCHWACHA, and N. KLECKNER, 1996  Synaptonemal complex (SC) component Zip1 plays a role in meiotic recombination independent of SC polymerization along the chromosomes. Proc. Natl. Acad. Sci. USA 93:9043-9048[Abstract/Free Full Text].

STUBER, C. W., J. F. WENDEL, M. M. GOODMAN and J. S. C. SMITH, 1988 Techniques and scoring procedures for starch electrophoresis of enzymes from maize (Zea mays L.). NC Agric. Res. Serv. Tech. Bull., 286.

SYBENGA, J., 1975 Meiotic Configurations. Springer-Verlag, New York.

TIMMERMANS, M. P., O. P. DAS, J. M. BRADEEN, and J. MESSING, 1997  Region-specific cis- and trans-acting factors contribute to genetic variability in meiotic recombination in maize. Genetics 146:1101-1113[Abstract].

WILLIAMS, C. G., M. M. GOODMAN, and C. W. STUBER, 1995  Comparative recombination distances among Zea mays L. inbreds, wide crosses and interspecific hybrids. Genetics 141:1573-1581[Abstract].