PrBn, a Major Gene Controlling Homeologous Pairing in Oilseed Rape (Brassica napus) Haploids

Corresponding author: Eric Jenczewski, INRA-GAP Rennes, Domaine de la Motte, BP 35327, F-35653 Le Rheu Cedex, France., jenczews{at}rennes.inra.fr (E-mail)

Communicating editor: P. J. PUKKILA

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Precise control of chromosome pairing is vital for conferring meiotic, and hence reproductive, stability in sexually reproducing polyploids. Apart from the Ph1 locus of wheat that suppresses homeologous pairing, little is known about the activity of genes that contribute to the cytological diploidization of allopolyploids. In oilseed rape (Brassica napus) haploids, the amount of chromosome pairing at metaphase I (MI) of meiosis varies depending on the varieties the haploids originate from. In this study, we combined a segregation analysis with a maximum-likelihood approach to demonstrate that this variation is genetically based and controlled mainly by a gene with a major effect. A total of 244 haploids were produced from F₁ hybrids between a high- and a low-pairing variety (at the haploid stage) and their meiotic behavior at MI was characterized. Likelihood-ratio statistics were used to demonstrate that the distribution of the number of univalents among these haploids was consistent with the segregation of a diallelic major gene, presumably in a background of polygenic variation. Our observations suggest that this gene, named PrBn, is different from Ph1 and could thus provide complementary information on the meiotic stabilization of chromosome pairing in allopolyploid species.

POLYPLOIDY has played a major role in the evolution of higher plants. Recent estimates suggest that up to 70% of all angiosperms have experienced one or more episodes of polyploidization during the course of their evolution (MASTERSON 1994 ) or domestication (HILU 1993 ; VAN RAAMSDONK 1995 ). Nonetheless, most if not all polyploids behave as diploids at meiosis, indicating that precise control of chromosome pairing is a prerequisite and confers evolutionary advantages in polyploid species. In allopolyploids containing homeologous chromosomes with sufficient homology to be able to pair at meiosis, cytological diploidization requires homeologous pairing to be suppressed. This process can be achieved by two complementary systems: (i) differentiation of homeologous chromosomes due to either structural changes or gene mutations, which leads to differential affinity and preferential pairing of homologs, and (ii) a genetic control that distinguishes between the differentiated sets of chromosomes and precludes pairing between homeologues. Although the presence of genetic systems regulating pairing has been suspected in a wide range of polyploids [e.g., cotton (KIMBER 1961 ), oat (GAUTHIER and MCGINNIS 1968 ), and fescues (JAUHAR 1975 )], evidence to date has been circumstantial. It is only in wheat that the presence of pairing regulators has been indisputably demonstrated with the characterization of the Ph1 locus (RILEY and CHAPMAN 1958 ; SEARS and OKAMOTO 1958 ) that suppresses homeologous pairing and contributes to the karyotypic stability of wheat (SANCHEZ-MORAN et al. 2001 ). Several other weaker loci have also been shown to either restrict or promote homeologous pairing (RILEY and LAW 1965 ). The question remains whether similar pairing regulators are widespread among polyploid species and have therefore a general evolutionary significance. Differences in meiotic pathways among closely related species (SHAW and MOORE 1998 ; CUNADO and SANTOS 1999 ) and in strategies for bivalent formation in different allopolyploids (JENKINS and REES 1991 ) indicate that the genetic mechanisms characterized in wheat are not the same as in other polyploid species. It is therefore necessary to explore new and complementary models to further understanding of polyploid meiotic diploidization. This issue has practical applications. A large number of successful alien introgressions have been achieved in wheat through homeologous recombination (FRIEBE et al. 1996 ) and, notably, by suppressing the control exerted by Ph1 (e.g., RILEY et al. 1968 ; LUO et al. 1996 and references therein; BENAVENTE et al. 2001 ). It may be anticipated that a better understanding of the genetic systems regulating homeologous pairing in other polyploid species could help in promoting and engineering introgressions.

Oilseed rape (Brassica napus) is an allopolyploid species (AACC; 2n = 38) that originated from hybridization between B. oleracea (CC; 2n = 18) and B. rapa (AA; 2n = 20). This species exhibits a clear bivalent-pairing regime and a disomic inheritance, which demonstrate that homologs pair at meiosis at the expense of homeologous pairing. The basis of such a diploid-like meiotic behavior is hypothetical. Different authors have proposed that homeologous pairing is genetically regulated in oilseed rape (ATTIA and ROBBELEN 1986A ; SHARPE et al. 1995 ) and its close relatives (PRAKASH 1974 ; HARDBERG 1976 ; EBER et al. 1994 ). RENARD and DOSBA 1980 and ATTIA and ROBBELEN 1986A observed that the amount of chromosome pairing in haploid plants (AC; n = 19) originating from different oilseed rape varieties was variable and identified high- and low-pairing varieties (at the haploid stage). Chiasmata were formed between paired chromosomes and resulted in both rod- and ring-shaped bivalents, but also in multivalents.

The objectives of this study are to determine if a large part of the variation observed for the amount of chromosome pairing in oilseed rape haploids is genetically based and to establish the genetic basis of this variation. To do so, we analyzed the meiotic behavior of haploids produced from a high- and a low-pairing line and developed proper statistical analyses to account for the different sources of variation (genetic and environmental determinants). We studied the segregation of the meiotic behavior in a population of haploids produced from F₁ hybrids between the high- and low-pairing lines and used likelihood-ratio (LR) statistics to test for alternative modes of inheritance and interpret genetic distribution in terms of both major and minor gene effects.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plant materials:
The genealogy and structure of the data sets are detailed in Fig 1. All haploids were isolated using microspore cultures as described by POLSONI et al. 1988 . A total of 13 and 27 haploids were isolated from a spring Korean line (Yudal) and a French dwarf winter line (Darmor-bzh), which are known to vary in their meiotic behavior at the haploid stage (Fig 1). All the diploid lines used to produce these parental haploids (Darmor-bzh F₃ and F₄ progenies and Yudal F₉ and F₁₃ progenies) were obtained by single-seed descent (SSD). These haploids comprise the parental data set. The parental genotypes differed in their response to produce haploids so that twice as many haploids were scored for Darmor-bzh as for Yudal. A total of 244 haploids were isolated from F₁ hybrids between Darmor-bzh and Yudal and comprise the offspring data set.

View larger version (25K): In this window In a new window Download PPT slide   Figure 1. Genealogy and structure of the parental (P) and offspring (O) subpopulations employed for segregation analyses. The total number of pollen mother cells (PMCs) observed is indicated for each series in each subpopulation. All the diploid lines used to produce the parental haploids were obtained by single-seed descent (SSD).

In an initial phase, three and seven haploids were produced from a few diploid plants from Yudal F₉ and Darmor-bzh F₃ progenies, respectively. A total of 45 haploids were isolated from F₁ hybrids obtained by crossing a single plant of the Darmor-bzh F₃ progeny to a single plant of the Yudal F₉ progeny (Fig 1). These parental and offspring haploids were grown together in the greenhouse and floral buds were sampled on almost the same date (three dates within 15 days); these haploids comprise the first set of observations (series 1). In a second phase, 10 and 20 haploids were isolated from Yudal F₁₃ and Darmor-bzh F₄ progenies. A total of 199 haploids were isolated from a few F₁ hybrids obtained by crossing a single Darmor-bzh F₄ plant by a single Yudal F₁₃ plant; three microspore cultures were needed to isolate all the haploids. Accordingly three series of haploids were successively grown in the greenhouse and analyzed separately (series 2–4 of observations). Only one set was grown simultaneously with parental haploids (series 3). For series 2 and 4, floral buds were sampled on three to four dates within 15 days. For series 3, floral buds were sampled on three dates within 1 month. For series 1 and 3, some haploids were observed at each date (or at least more than once) and showed the same meiotic behavior (data not shown). Sixteen other haploids were observed in 2 consecutive years to test for the repeatability of the amount of pairing. These haploids were chosen within series 2–4 to encompass the whole range of meiotic behaviors [3.3 < no. of univalents < 10]. These haploids were conserved as cuttings.

Meiotic observations:
Floral buds were fixed in Carnoy's solution (ethanol-chloroform-acetic acid, 6:3:1) for 24 hr and stored in 50% ethanol. Observations on the pollen mother cells (PMCs) were performed at the metaphase I (MI) stage from anthers squashed and stained in a drop of 1% acetocarmine solution. On average, 20 PMCs (minimum, 14; maximum, 149) were examined for each haploid, regardless of their origin.

Statistical analysis:
Statistical analyses were performed mainly on the number of univalents. This variable was chosen because it can be reliably scored and because it measures the whole extent of pairing in a synthetic way, reflecting by subtraction the number of chromosomes associated as both bivalents and multivalents. Parental data were first analyzed on their own, to determine to what extent variation in the amount of pairing among Darmor-bzh and Yudal haploids was genotypically determined. On the basis of this preliminary analysis, the offspring and parental data were then analyzed simultaneously, so that parental and offspring distributions could be compared within a single model. In the models below, we denote by Y_g,lij the number of univalents in the PMC j of haploid i observed in the series l from population g, where g refers to haploids produced from Darmor-bzh (g = D), Yudal (g = Y), or Darmor-bzh x Yudal F₁ hybrids (g = H).

Analysis of parental data:
The model employed for each parental genotype was

(1)

where µ_g is the mean for population g (g is either genotype D or genotype Y), _l is the effect of series l (l = 1 or 3), b_g,li is a random haploid plant effect, and _g,lij is a residual error term. The b_g,li and _g,lij random effects were assumed to follow independent normal and centered distributions, with variances denoted by ²_g for haploid effects and by ²_g for residual errors. The parameter estimates in model (1) and their asymptotic standard errors were calculated by residual maximum likelihood (REML) with the PROC MIXED procedure of SAS (SAS INSTITUTE 1999). Note that for variance parameter estimates, the standard errors are known to be unreliable and should not be used to construct confidence intervals. The hypotheses on the absence of series effects (₁ = ₃ = 0 vs. _{1 3}) and of haploid effects (²_g = 0 vs. ²_g > 0) were tested by an analysis of variance performed with the PROC GLM procedure of SAS. The RANDOM statement of this procedure was used because the haploid random factor was nested within the series factor. On two occasions in RESULTS, we propose to quantify and compare the contributions to the variability between haploids that are due to the different factors of the models. For factors with random effects, contributions are given by the estimated variance parameters. For factors with fixed effects, contributions are calculated as the average squared estimates of the factor effects.

Simultaneous analysis of parental and offspring data:
We considered a model with both a segregating diallelic major gene and a completely additive polygenic background. This model allowed the presence of two subpopulations in the offspring data set, one with a behavior similar to Darmor-bzh haploids (denoted HD) and the other one with a behavior similar to Yudal haploids (denoted HY). Our general model was

(2)

where µ_D, µ_Y, _l, b_D,li, b_Y,li, _D,lij, and _Y,lij are as defined for the parental data; µ_HD and µ_HY are the means for the two subpopulations HD and HY; b_HD,li and b_HY,li are the random haploid plant effects for the two subpopulations HD and HY; _H,lij is a residual error term; and p and 1 - p are the transmission probabilities of the Darmor-bzh and Yudal major-locus alleles, respectively. The b_g,li and _g,li random effects were assumed to follow independent normal and centered distributions. Haploid variances ²_g (with g = D, Y, HD, or HY) were assumed to depend on the genotype, whereas the residual variance ² was assumed to be the same for all PMCs. Model (2) involves factors with fixed (genotype and series) and random effects (plants) and is therefore a mixed model. In the offspring data, it includes the mixture of two distributions with different means and variances (EVERITT and HAND 1981 ).

The vector of parameters is = (p, ₁, ₂, ₃, ₄, µ_D, µ_Y, µ_HD, µ_HY, ²_D, ²_Y, ²_HD, ²_HY, ²), with the constraints ₁ + ₂ + ₃ + ₄ = 0 to avoid overparameterization and µ_HY - µ_HD > 0 to ensure that the parameters can be identified. The parameter estimates and their asymptotic standard errors were calculated by Gaussian likelihood maximization. For each g, l, i (g = D, Y, or H), Y_g,li denotes the vector of observations on all PMCs from plant i of genotype g in series l. In the model, the Y_g,li's are mutually independent and so the total log-likelihood is equal to the sum of the log-likelihoods for each vector Y_g,li. (Y;m,) denotes the density of a Gaussian vector with expectation m and covariance matrix , calculated in Y. For g = D or Y, the likelihood for Y_g,li is (Y_g,li;(₁ + µ_g)1, ²_gJ + ²I), where 1 denotes the vector of ones of appropriate length, I is the identity matrix, and J is the matrix of ones. For g = H, the likelihood for Y_g,li is p(Y_g,li;(_l + µ_HD)1, ²_HDJ + ²I) + (1 - p)(Y_g,li;(_l + µ_HY)1, ²_HYJ + ²I), which we denote by f(Y_g,li;).

The maximization of the likelihood coming from a mixture model is usually carried out using an expectation-maximization (EM) algorithm (DEMPSTER et al. 1977 ). In this study, the numerical procedure was improved by using a generalization of this algorithm, namely an expectation-conditional-maximization (ECM) algorithm (MENG and RUBIN 1993 ). Indeed, the ECM algorithm allowed us to maximize the log-likelihood more efficiently, by separating maximization with respect to the variance parameters and maximization with respect to the other model parameters. The algorithm was programmed using the MATLAB software (MATHWORKS 2000 ).

The testing procedure was based on the LR test statistic. This procedure tests the null hypothesis that the vector of parameters satisfies a set of q linear constraints against the alternative that at least one of these constraints is not satisfied. LR equals twice the difference between the maximum log-likelihoods under the alternative and null hypotheses. The null hypothesis is rejected at level 5% when the test statistic is greater than the 95% quantile of a ² distribution with q d.f. This test proved to be approximately of level 5% when the numbers of haploids and PMCs per haploid are large (GRAYBILL 1976 ).

Finally, we used the estimated model parameters to predict the major-locus genotype of all haploids in the offspring data set. According to Bayes' theorem, the probability that a haploid from the offspring data set carries the Darmor-bzh allele, conditionally to its vector Y_H,li of observed values, is

and the probability that it carries the Yudal allele is P(Y/Y_H,li) = 1 - P(D/Y_H,li) (EVERITT and HAND 1981 ).

For estimating the repeatability of our observations, we computed the correlation between repeated measures (mean number of univalents) on the 16 haploids from the offspring data set that have been observed in 2 consecutive years (FALCONER and MACKAY 1996 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Analysis of the variation among Darmor-bzh and Yudal haploids:
Typical pairing patterns at MI for Darmor-bzh and Yudal haploids are illustrated in Fig 2. Fig 3A presents the mean numbers of univalents for each plant in the parental data set. Averaged meiotic behaviors for Darmor-bzh and Yudal haploids estimated by REML and corrected for the series and haploid effects (Table 1) demonstrate that pairing patterns in Darmor-bzh and Yudal haploids are clear cut. Haploids produced from Darmor-bzh showed far more pairing than those originating from Yudal; 80% of the PMCs observed in the Darmor-bzh haploids had less than six univalents whereas 95% of the PMCs scored in the Yudal haploids had more than eight univalents. On average, only 36.8% of the chromosomes paired in the Yudal haploids while >75% of the chromosomes were associated in the Darmor-bzh haploids. Similar differences were observed with the number of multivalents: 41 trivalents (III) and 60 quadrivalents (IV) were scored in a total of 593 PMCs from 27 Darmor-bzh haploids, whereas 19 III and only 2 IV were scored over the 317 PMCs analyzed from the 13 Yudal haploids. Regardless of the genotype, bivalents and multivalents were held by chiasmata.

View larger version (317K): In this window In a new window Download PPT slide   Figure 2. First metaphase of meiosis in 1% acetocarmine-stained squashes of pollen mother cells of two oilseed rape haploids produced from the parental lines: (A) low pairing in a Yudal haploid, with two bivalents and 15 univalents; (B) high pairing in a Darmor-bzh haploid, with eight bivalents and only 3 univalents.

View larger version (24K): In this window In a new window Download PPT slide   Figure 3. Mean numbers of univalents per PMC for each haploid produced from either (A) the parental lines or (B) the hybrids between Darmor-bzh and Yudal. Haploids produced from Darmor-bzh are represented by solid histograms and those produced from Yudal by open histograms. Asterisk (*) points to haploids with intermediate behavior.

Using the number of univalents as a variable, the estimated values (plus or minus their standard errors) for the parameters of model (1) were µ_D = 4.82 (±0.08), ₁ = -₃ = 0.98 (±0.08), ²_D = 0.02 (±0.04), ² = 2.52 (±0.15) for Darmor-bzh haploids and µ_Y = 12.03 (±0.44), ₃ = -₁ = 0.26 (±0.33), ²_Y = 0.86 (±0.44), ² = 3.78 (±0.31) for Yudal haploids, respectively. The analyses of variance (Table 2), performed separately on each parental line, showed that significant differences existed between the two series of haploids (series 1 and 3) produced from Darmor-bzh (P = 1.8e-11), which differed on average by the association of two chromosomes as a bivalent. By contrast, no differences were observed between the two series of haploids produced from Yudal whereas haploids within each series were significantly different from one another (P = 2.2e-09). This result is surprising because the diploid Yudal plants used for microspore culture were from the same F₉ or F₁₃ progenies and were therefore genetically almost homogeneous. By contrast, no differences were detected among the haploids of the same generation produced from Darmor-bzh.

According to parameter estimates, 93% of the observed variability for the number of univalents could be attributed to differences between the parental genotypes, 4% to differences between series (calculated by averaging over the two parental lines), and 3% to differences between haploids within a series.

Analysis of the whole data set, including the segregating population of haploids:
Fig 3B presents the mean numbers of univalents for each plant in the offspring data set. These values proved to be very repeatable and reliable using Falconer's method; a very high correlation (r_F = 0.96) was observed among repeated measures on the 16 haploids that had been chosen in the offspring data set to encompass the whole range of meiotic behaviors. The maximum absolute differences in the mean number of univalents between the 2 years of observation for a plant were 1.7 and then 1.05.

The distribution of the mean number of univalents in the offspring data set was clearly bimodal with a mixture of two distinct distributions (Fig 4). According to model (2), the parameter estimates (plus or minus their standard errors) were p = 0.46 (±0.03), µ_D = 4.50 (±0.36), µ_Y = 12.06 (±0.38), µ_HD = 4.67 (±0.35), µ_HY = 10.27 (±0.60), ₁ = 1.08 (±0.36), ₂ = -0.05 (±0.27), ₃ = -0.56 (±0.35), ₄ = -0.48 (±0.62), ²_D = 0.005 (±0.07), ²_Y = 0.97 (±1.05), ²_HD = 0.70 (±0.21), ²_HY = 0.91 (±1.09), and ² = 3.11 (±0.17). Adjusted and observed distributions of the mean number of univalents for each haploid of the offspring data set were in close agreement (Fig 4).

View larger version (43K): In this window In a new window Download PPT slide   Figure 4. Comparison of the observed (histogram) and estimated (solid curve) frequency distributions of the mean number of univalents in the offspring haploids. The observed distribution was obtained by pooling observations (mean number of univalents, adjusted for the estimated series effects) from all the series of the offspring data set. Asterisks (*) point to haploids with intermediate behavior.

We initially tested whether the distribution of the number of univalents was consistent with the Mendelian segregation of a major gene. The full model that treated transmission probability as an unknown parameter was compared, using a likelihood-ratio test, with the restricted model that fixed p = 0.5. As the full model did not provide a better fit than the restricted one (P = 0.25), the hypothesis p = 0.5 was accepted at the 5% level (Table 3). Therefore the distribution of the mean number of univalents in the offspring data set supports the presence of a major gene with two alleles.

To analyze whether this major gene was the only genetic source of variation, we tested whether the distribution of the number of univalents in the offspring data set was consistent with the mixture of the two parental distributions within a given series: this would be expected if the amount of pairing was completely controlled by the major gene. We compared the full model that treated µ_HD, µ_HY, ²_HD, and ²_HY as free parameters with restricted models that fixed µ_D = µ_HD, µ_Y = µ_HY, ²_HD = ²_D, or ²_HY = ²_Y. Note that, as a consequence of using a mixture model, each test was performed on the whole data set and took account of the uncertainty on the subpopulations of offspring haploids. Results, presented in Table 3, clearly showed that the distribution of the number of univalents in the offspring data set was not a mixture of the two parental distributions. First, the mean number of univalents was significantly lower in the HY subpopulation than in the parental Yudal haploids. By contrast there were no differences between the Darmor-bzh mean and that of the HD subpopulation (P = 0.17), although µ_HD was slightly higher than µ_D. Second, although ²_HY was not significantly different from ²_Y, the variance for the number of univalents was significantly higher in the HD subpopulation than in the parental D subpopulation; actually, ²_Y, ²_HY, and ²_HD were not different from one another (Table 3).

According to the parameter estimates of model (2), 86% of the observed variability for the number of univalents in the offspring data set was due to differences between the HD and HY subpopulations, 5% to differences between series, and 9% to differences between haploids within a series and a subpopulation. Interestingly, several haploids in the offspring data set exhibited an intermediate pairing behavior (Fig 3B and Fig 4). Prediction of the major-locus genotype of each haploid in the offspring data set showed that three of them had probabilities P(D/Y_H,li) and P(Y/Y_H,li) <0.9. These three plants had an averaged meiotic behavior of 6.77 univalents (I) + 5.97 bivalents (II) + 0.025 III + 0.05 IV; 90% of their PMCs had 5 I + 7 II, 7 I + 6 II, or 9 I + 5 II. These patterns were quite different from those observed in the other haploids of the offspring data set since 82% of the PMCs had more than nine univalents in the HY subpopulation or less than five univalents in the HD subpopulation. The amount of pairing in two of these intermediate haploids was measured twice and values were similar in both measurements (data not shown).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Several authors have observed a variation in the extent of pairing among oilseed rape haploids (RENARD and DOSBA 1980 ; ATTIA and ROBBELEN 1986A ). In this study, we demonstrated that this variation is genetically based and controlled mainly by the major gene PrBn (Pairing regulator in B. napus).

Investigations on the meiotic behavior of poly-haploids have been made on a wide range of allopolyploid species, usually at MI or later stages (KIMBER and RILEY 1963 ; MAGOON and KHANNA 1963 ). These studies have usually demonstrated a low level of bivalent formation and/or various types of secondary associations of univalents (i.e., not held by chiasmata). In this study, we observed a high level of pairing in the oilseed rape haploids isolated from Darmor-bzh, with up to 75% of the chromosomes being associated at MI. Even in the low-pairing haploids isolated from Yudal, a minimum of two to three bivalents were systematically observed. All these associations were held by chiasmata and probably resulted from both auto- and allosyndesis within and between the A and C genomes of oilseed rape. Autosyndesis, the pairing between two chromosomes originating from the same genome, has been reported within B. oleracea (ARMSTRONG and KELLER 1982 ) and B. rapa (ARMSTRONG and KELLER 1981 ) as a result of intragenomic duplications (PRAKASH and HINATA 1980 ; SCHMIDT et al. 2001 ). As only one or two autosyndetic pairs are possible within the C and A genomes (MIZUSHIMA 1950 , MIZUSHIMA 1972 ; ARMSTRONG and KELLER 1981 , ARMSTRONG and KELLER 1982 ), additional associations should be considered as allosyndetic (i.e., involving chromosomes from different genomes). This assertion is consistent with the close proximity of A/C homeologous genomes of oilseed rape and with their high affinity for pairing, which is indirectly supported by the high amount of pairing reported in B. rapa x B. oleracea interspecific hybrids (ATTIA and ROBBELEN 1986B ); for example, INOMATA 1980 observed that 81% of the PMCs in such a hybrid contained 1 I + 9 II, 8 II + 1 III, 8 II + 3 I, or 2 I + 7 II + 1 III and that the frequency of quadrivalents was 10%.

Interestingly, high-pairing haploids of oilseed rape exhibit a meiotic behavior that is almost similar to that of raw B. rapa x B. oleracea interspecific hybrids (direct comparisons are ongoing). This suggests that these haploids express the largest extent of pairing affinities between the A and C genomes. By contrast, low-pairing haploids show a severe restriction in pairing potentialities. Such restriction has been used to infer the presence of pairing regulators (KIMBER 1961 ; RILEY and LAW 1965 ) or interpreted as the consequence of an overdifferentiation of homeologous chromosomes since the origin of the polyploid state. This last proposal seems unlikely in oilseed rape haploids for several reasons. First, meiosis in Darmor-bzh x Yudal F₁ diploid hybrids is regular (19 II; data not shown), suggesting that these genotypes do not differ by extensive chromosomal rearrangements. In addition, high levels of chromosome pairing in A x AC and AC x C hybrids indicated that the A/C genomes in oilseed rape have remained essentially unaltered with respect to the A/C genomes of their progenitors (OLSSON and HAGBERG 1955 ; ATTIA et al. 1987 ; see also PARKIN et al. 1995 ). Finally our study provided direct evidence that the differences between the high- and low-pairing haploids are genetically based.

Our study combined a segregation analysis with a maximum-likelihood approach to test for different modes of inheritance of the pattern of chromosome pairing in oilseed rape haploids. A similar approach has been recently advocated by WU et al. 2001 to combine quantitative genetic and population genetic principles. Our approach assumed normality of the underlying distributions, which appeared consistent with the observed mean numbers of univalents and simplified the form of the likelihood functions. Our statistical treatment provides a powerful and flexible framework to investigate the different sources of variation, test for parameter adjustment in the different parental (D, Y) and offspring (HD, HY) subpopulations independently, and interpret genetic data in terms of both major and minor gene segregation.

Segregation analysis combined with LR tests clearly demonstrates that pairing patterns in oilseed rape haploids are inherited in a Mendelian fashion and supports the presence of a single major gene. However, the distribution of the number of univalents in the offspring data set was not consistent with the mixture of the two parental distributions; an obvious asymmetry in the evolution of mean and variance parameters in the HD and HY subpopulations (i.e., µ_D = µ_HD and ²_HD > ²_D while µ_Y > µ_HY and ²_HY = ²_Y) was detected. This pattern may have resulted from the segregation of additional weaker genes with nonadditive effects that are confounded with the major gene activity or the range of chromosome pairing affinities, environmental variation affecting HD and HY haploids in a different way, or both. These interpretations are tentative although they are supported by additional observations. On the one hand, the meiotic behavior of Yudal haploids, which were taken and observed at the same date, was related to their position in the greenhouse (data not shown); this indicates that a large part of the unexpected variation observed between these haploids (Table 2) was due to environmental heterogeneity. By contrast, no variation was detected among Darmor-bzh haploids (Table 2), suggesting that pairing in high-pairing haploids was less susceptible to environmental variation than pairing in low-pairing haploids. On the other hand, strong differences between the Y and HY subpopulations (Table 3), the presence of intermediate haploids with a repeatable behavior, and the increased variance in the HD subpopulations (while pairing patterns in Darmor-bzh haploids did not vary) are consistent with the presence of a polygenic background.

Our results suggest that control of chromosome pairing in oilseed rape haploids is roughly similar to that in wheat in that major genes are involved in both cases. However, it is likely that PrBn is different from Ph1. First, polymorphism observed among oilseed rape haploids is natural whereas there is hardly any natural polymorphism for Ph1 (see OZKAN and FELDMAN 2001 ); the only known wheat lines defective for Ph1 have been induced through irradiation (SEARS 1977 ; GIORGI 1978 ; ROBERTS et al. 1999 ). Second, Ph1 prevents homeologous pairing at both the haploid and diploid stage. By contrast, all B. napus accessions, regardless of the frequency of chromosome pairing in their dihaploid forms, display regular bivalent associations and disomic inheritance. This indicates that, if the presence of Ph1 is essential for chromosome stability and fertility in wheat (SANCHEZ-MORAN et al. 2001 ), PrBn is not required. Alternatively, PrBn could contribute to the regularity of chromosome pairing in all diploid forms of B. napus, but the allele present in genotypes with a high-pairing behavior at the haploid stage could be ineffective at the hemizygous stage or at least less efficient than that in the diploid state. Such haplo-ineffective regulating systems have been described in hexaploid fescues (JAUHAR 1975 ) and different Aegilops species (CUNADO and SANTOS 1999 ).

This last hypothesis is tentative and clearly deserves further examination. Ongoing genetic mapping and subsequent cloning of PrBn, comparative analysis of chromosome pairing at prophase I in high- and low-pairing haploids, and direct studies of the amount of recombination in oilseed rape diploids and haploids should further our understanding of the genetic regulation of chromosome pairing in this species. Then, combined with the extensive and continuous characterization of Ph1 (ROBERTS et al. 1999 ), the Brassica model could provide new insights into the nature of the meiotic stabilization of allopolyploid species.

	ACKNOWLEDGMENTS

We thank J. C. Letanneur and D. Simmoneaux for technical assistance, Dr. E. Klein for his help in statistical analysis, and Drs. K. Alix, R. Delourme, and D. Gaudet for their fruitful comments on the manuscript.

Manuscript received July 23, 2002; Accepted for publication February 4, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ARMSTRONG, K. C. and W. A. KELLER, 1981  Chromosome pairing in haploids of Brassica campestris.. Theor. Appl. Genet. 59:49-52.

ARMSTRONG, K. C. and W. A. KELLER, 1982  Chromosome pairing in haploids of Brassica oleracea.. Can. J. Genet. Cytol. 24:735-739.

ATTIA, T. and G. RÖBBELEN, 1986a  Meiotic pairing in haploids and amphihaploids of spontaneous versus synthetic origin in rape, Brassica napus L. Can. J. Genet. Cytol. 28:330-334.

ATTIA, T. and G. RÖBBELEN, 1986b  Cytogenetic relationship within cultivated Brassica analyzed in amphihaploids from three diploid ancestors. Can. J. Genet. Cytol. 28:323-329.

ATTIA, T., C. BUSSO, and G. RÖBBELEN, 1987  Digenomic triploids for an assessment of chromosome relationships in the cultivated diploid Brassica species. Genome 29:326-330.

BENAVENTE, E., K. ALIX, J. C. DUSAUTOIR, J. ORELLANA, and J. L. DAVID, 2001  Early evolution of the chromosomal structure of Triticum turgidum-Aegilops ovata amphiploids carrying and lacking the Ph1 gene. Theor. Appl. Genet. 103:1123-1128.

CUÑADO, N. and J. L. SANTOS, 1999  On the diploidization mechanism of the genus Aegilops: meiotic behaviour of interspecific hybrids. Theor. Appl. Genet. 99:1080-1086.

DEMPSTER, A., N. M. LAIRD, and D. B. RUBIN, 1977  Maximum likelihood estimation from incomplete data via the EM algorithm. J. R. Stat. Soc. B 39:1-38.

EBER, F., A. M. CHÈVRE, A. BARANGER, P. VALLÉE, and X. TANGUY et al., 1994  Spontaneous hybridization between a male-sterile oilseed rape and two weeds. Theor. Appl. Genet. 88:362-368.

EVERITT, B. S., and D. J. HAND, 1981 Finite Mixture Distributions. Chapman & Hall, London.

FALCONER, D. S., and T. F. C. MACKAY, 1996 Introduction to Quantitative Genetics, Ed. 4. Longman Science and Technology, Harlow, UK.

FRIEBE, B., J. JIANG, W. J. RAUPP, R. A. MCINTOSH, and B. S. GILL, 1996  Characterization of wheat-alien translocations conferring resistance to diseases and pests: current status. Euphytica 91:59-87.

GAUTHIER, F. M. and R. C. MCGINNIS, 1968  The meiotic behaviour of a nulli-haploid plant in Avena sativa L. Can. J. Genet. Cytol. 10:186-189.

GIORGI, B., 1978  A homoeologous pairing mutant isolated in Triticum durum cv. Capelli. Mut. Breed. Newsl. 11:4-5.

GRAYBILL, F. A., 1976 Theory and Application of the Linear Model. Duxbury Press, North Scituate, MA.

HARDBERG, D. J., 1976 Cytotaxonomic studies of Brassica and related genera, pp. 47–68 in The Biology and Chemistry of the Crucifereae, edited by J. G. VAUGHAN, A. J. MACLEOD and B. M. JONES. Academic Press, London.

HILU, K. W., 1993  Polyploidy and the evolution of domesticated plants. Am. J. Bot. 80:1494-1499.

INOMATA, N., 1980  Hybrid progenies of the cross Brassica campestris x Brassica oleracea. I. Cytogenetical studies on F1 hybrids. Jpn. J. Genet. 55:189-202.

JAUHAR, P. P., 1975  Genetic regulation of diploid-like chromosome pairing in the hexaploid species, Festuca arundinacea Schreb. and F. rubra L. (Gramineae). Chromosoma 52:363-382.

JENKINS, G. and H. REES, 1991  Strategies of bivalent formation in allopolyploid plants. Proc. R. Soc. Lond. Ser. B 243:209-214.

KIMBER, G., 1961  Basis of the diploid-like meiotic behaviour of polyploid cotton. Nature 191:98-100.

KIMBER, G. and R. RILEY, 1963  Haploid angiosperm. Bot. Rev. 29(4):480-531.

LUO, M. C., J. DUBCOVSKY, S. GOYAL, and J. DVORAK, 1996  Engineering of interstitial foreign chromosome segments containing the K+/Na+ selectivity gene Kna1 by sequential homoeologous recombination in durum wheat. Theor. Appl. Genet. 93:1180-1184.

MAGOON, M. L. and K. R. KHANNA, 1963  Haploids. Caryologia 16:191-234.

MASTERSON, J., 1994  Stomatal size in fossil plants: evidence for polyploidy in majority of angiosperms. Science 264:421-424.[Abstract/Free Full Text]

MATHWORKS, 2000 Using MATLAB. The MathWorks, Natick, MA.

MENG, X. L. and D. B. RUBIN, 1993  Maximum likelihood estimation via the ECM algorithm: a general framework. Biometrika 80:267-278.[Abstract/Free Full Text]

MIZUSHIMA, U., 1950  Karyogenetic studies of species and genus hybrids in the tribe Brassicaceae of Cruciferae. Tohoku J. Agric. Res. 1:1-14.

MIZUSHIMA, U., 1972  Evolution of species in Brassiceae and their breeding. Kagaku Seibutsu 10:78-85.

OLSSON, G. and A. HAGBERG, 1955  Investigations on haploid rape. Hereditas 41:227-237.

OZKAN, H. and M. FELDMAN, 2001  Genotypic variation in tetraploid wheat affecting homoeologous pairing in hybrids with Aegilops peregrina.. Genome 44:1000-1006.[Medline]

PARKIN, I. A. P., A. G. SHARPE, D. J. KEITH, and D. J. LYDIATE, 1995  Identification of the A and C genomes of amphidiploid Brassica napus (oilseed rape). Genome 38:1122-1131.

POLSONI, L., S. KOTT, and W. D. BEVERSDORF, 1988  Large-scale microspore culture technique for mutation-selection studies in Brassica napus.. Can. J. Bot. 66:1681-1685.

PRAKASH, S., 1974  Probable basis of diploidization of Brassica juncea Coss. Can. J. Genet. Cytol. 16:232-234.

PRAKASH, S. and K. HINATA, 1980  Taxonomy, cytogenetics and origin of crop Brassica, a review. Opera Bot. 55:1-57.

RENARD, M. and F. DOSBA, 1980  Etude de l'haploidie chez le colza (Brassica napus L. var oleifera Metzger). Ann. Amel. Pl. 30:191-209.

RILEY, R. and V. CHAPMAN, 1958  Genetic control of the cytologically diploid behaviour of hexaploid wheat. Nature 13:713-715.

RILEY, R. and C. N. LAW, 1965  Genetic variation in chromosome pairing. Adv. Genet. 13:57-114.

RILEY, R., V. CHAPMAN, and R. JOHNSTON, 1968  Introduction of yellow rust resistance of Aegilops comosa into wheat by genetically induced homoeologous recombination. Nature 217:383-384.

ROBERTS, M. A., S. M. READER, C. DALGLIESH, T. E. MILLER, and T. N. FOOTE et al., 1999  Induction and characterization of Ph1 wheat mutants. Genetics 153:1909-1918.[Abstract/Free Full Text]

SÁNCHEZ-MORÁN, E., E. BENAVENTE, and J. ORELLANA, 2001  Analysis of karyotypic stability of homoeologous-pairing (ph) mutants in allopolyploid wheat. Chromosoma 110:371-377.[Medline]

SAS INSTITUTE, 1999 SAS/STAT User's Guide, Version 8. SAS Institute, Cary, NC.

SEARS, E. R., 1977  An induced mutant with homoeologous pairing in common wheat. Can. J. Genet. Cytol. 19:585-593.

SEARS, E. R. and M. OKAMOTO, 1958  Intergenomic chromosome relationships in hexaploid wheat. 10th Int. Congr. Genet. 2:258-259.

SHARPE, A. G., I. A. P. PARKIN, D. J. KEITH, and D. J. LYDIATE, 1995  Frequent non-reciprocal translocations in the amphidiploid genome of oilseed rape (Brassica napus). Genome 38:1112-1121.

SHAW, P. and G. MOORE, 1998  Meiosis: vive la difference!. Curr. Opin. Plant Biol. 1:458-462.[Medline]

SCHMIDT, R., A. ACARKAN, and K. BOIVIN, 2001  Comparative structural genomics in the Brassicaceae family. Plant Physiol. Biochem. 39:253-262.

VAN RAAMSDONK, L. W. D., 1995  The cytological and genetical mechanisms of plant domestication exemplified by four crop models. Bot. Rev. 61:367-399.

WU, R., B. LI, S. S. WU, and G. CASELLA, 2001  A maximum likelihood-based method for mining major genes affecting a quantitative character. Biometrics 57:764-768.[Medline]

This article has been cited by other articles:

				R. T. Gaeta, J. C. Pires, F. Iniguez-Luy, E. Leon, and T. C. Osborn Genomic Changes in Resynthesized Brassica napus and Their Effect on Gene Expression and Phenotype PLANT CELL, November 1, 2007; 19(11): 3403 - 3417. [Abstract] [Full Text] [PDF]

				S. D. Nicolas, G. L. Mignon, F. Eber, O. Coriton, H. Monod, V. Clouet, V. Huteau, A. Lostanlen, R. Delourme, B. Chalhoub, et al. Homeologous Recombination Plays a Major Role in Chromosome Rearrangements That Occur During Meiosis of Brassica napus Haploids Genetics, February 1, 2007; 175(2): 487 - 503. [Abstract] [Full Text] [PDF]

				Z. Liu, K. Adamczyk, M. Manzanares-Dauleux, F. Eber, M.-O. Lucas, R. Delourme, A. M. Chevre, and E. Jenczewski Mapping PrBn and Other Quantitative Trait Loci Responsible for the Control of Homeologous Chromosome Pairing in Oilseed Rape (Brassica napus L.) Haploids Genetics, November 1, 2006; 174(3): 1583 - 1596. [Abstract] [Full Text] [PDF]

				I. A. P. Parkin, S. M. Gulden, A. G. Sharpe, L. Lukens, M. Trick, T. C. Osborn, and D. J. Lydiate Segmental Structure of the Brassica napus Genome Based on Comparative Analysis With Arabidopsis thaliana Genetics, October 1, 2005; 171(2): 765 - 781. [Abstract] [Full Text] [PDF]

				J. A. Udall, P. A. Quijada, and T. C. Osborn Detection of Chromosomal Rearrangements Derived From Homeologous Recombination in Four Mapping Populations of Brassica napus L. Genetics, February 1, 2005; 169(2): 967 - 979. [Abstract] [Full Text] [PDF]