Cytogenetic map of common bean (Phaseolus vulgaris L.)

In this article, we present the mapping of the remaining eight chromosome pairs of common bean using 20 clones that gave unique sequences with or without blocking DNA, complementing the previous mapping of chromosomes 3, 4, and 7 (Pedrosa-Harand et al. 2009). We also propose a set of four BAC clones that can be used to recognize all chromosomes of the complement of BAT93, aiding future mapping. This is the third cytogenetic map of a legume species, after L. japonicus (Pedrosa et al. 2002) and Medicago truncatula (Kulikova et al. 2001), and the first for a tropical legume. Because of its proximity to other economically important species from the phaseoloids clade such as soybean and cowpea (Gepts et al. 2005), it will be useful for further syntenic studies in the group, especially considering the tetraploid nature of soybean (Shoemaker et al. 2006; Schlueter et al. 2008).

The common bean cytogenetic map presented here is integrated into the core linkage map of the species (Freyre et al. 1998) through BACs selected using markers from the Florida map (Vallejos et al. 1992) and the recently developed map using comparative anchor-tagged sequence loci (Hougaard et al. 2008). Except for chromosome 5, for which only one anchoring point could be established, LGs were assigned and oriented in respect to chromosome short and long arms, confirming the orientation of LGs and chromosomes recently proposed (Pedrosa-Harand et al. 2008). LG I (Vallejos et al. 1992) corresponds to the short arm and part of the long arm of chromosome 10 only because the BAC corresponding to the terminal Bng200 marker mapped to the middle of the long arm. Although the long arm has a large terminal 45S rDNA cluster, the position of the clones mapped to this chromosome indicates that LG I does not have good coverage. Indeed, more terminal markers were mapped to the corresponding LG 10 (Hougaard et al. 2008), including an rDNA locus (Freyre et al. 1998) and an RAPD marker PROD15⁶⁸⁰ (Geffroy et al. 2000), in the vicinity of which the bacteriophage SJ19.12 was mapped. LGs G and K, on the other hand, apparently only correspond to the long arms of chromosomes 6 and 9, respectively. In both cases, short arms seem to be composed mainly of 45S rDNA. The 45S rDNA locus on chromosome 6 has been observed in all accessions of P. vulgaris investigated so far (Moscone et al. 1999; Pedrosa et al. 2003; Pedrosa-Harand et al. 2006). Variations in the size of this cluster were correlated with the arm and chromosome sizes. Indeed, although chromosome 6 is acrocentric and is the smallest chromosome in ‘BAT93,’ it is metacentric and the largest in ‘Saxa’ (Moscone et al. 1999; present results). Chromosome 9 also seems to carry a conserved rDNA locus (Pedrosa-Harand et al. 2006), but its morphology had not been correctly established before. It was initially considered to be metacentric (Moscone et al. 1999; Pedrosa et al. 2003), but its centromere was probably misidentified due to lack of a conspicuous primary constriction. The presence of pericentromeric repetitive sequences adjacent to the terminal rDNA cluster and the occasional observation after FISH of the DAPI staining of positive dots in the same region indicate that this chromosome is submetacentric in ‘BAT93’ and probably in all accessions of the species.

The order of markers in the LG was in complete agreement with the order of the corresponding BACs along the chromosomes. For chromosome 1, two BAC clones 38C24 and 257L12, corresponding to Bng171 and Bng173, respectively, colocalized on mitotic chromosomes, but could be ordered after mapping on pachytene chromosomes. BAC 257L12 mapped more terminally than 38C24, corresponding to the more terminal position of Bng173 relative to Bng171 in the LG. These BACs were adjacent to each other in pachytene chromosomes, confirming their close physical proximity and explaining the difficulty in ordering these markers with high confidence on the genetic map (Vallejos et al. 1992). The map of chromosome 11 confirmed the suppression of recombination in extended pericentromeric chromosome regions observed previously (Pedrosa-Harand et al. 2009): less than a quarter of the LG length corresponded to more than half of the chromosome size. The mapping of heterochromatin along chromosomes indicated that suppression of recombination correlates with the presence of prominent pericentromeric heterochromatic blocks (see Fig. 5).

The pericentromeric heterochromatin was defined by the repetitive distribution pattern of BAC 12M3, which gave pericentromeric signals, and the colocalization of these regions with the bright CMA⁺ bands after CMA/DAPI staining (see also Zheng et al. 1993). The BAC clone that mapped more proximally on the long arm of chromosome 11, BAC 25D1, and two other BACs (36H21 and 173P6) mapped to pericentromeric heterochromatin and needed the addition of C₀t − 100 blocking DNA to give single-copy signals. Blocking was, however, also necessary for some BACs that mapped to euchromatin. The occurrence of many BACs containing repetitive sequences, although all had been selected with single-copy markers, indicates that unique sequences are frequently interspersed with repetitive sequences, confirming the less compartmentalized nature of the common bean genome (Pedrosa-Harand et al. 2009). In complete agreement with this hypothesis, sequencing of the B4 resistance gene cluster in BAT93, combined with FISH experiments, revealed that low-copy genes were interspersed with repetitive sequences (Geffroy et al. 2009; David et al. 2010).

The high proportion of repetitive sequences intermingled with single-copy sequences in euchromatin and heterochromatin is not the only prominent feature of the relatively small common bean genome (approximately 600 Mb; Arumuganathan and Earle 1991). Half (48%) of its genome was estimated to be heterochromatic based on the proportion of rDNA (5% of the genome), subtelomeric blocks (9%), and pericentromeric blocks (34%) on mitotic chromosomes. This value may be underestimated because small regional heterochromatic regions are not detectable by the cytological procedures used (Houben et al. 2003), but it correlates well with estimates of the proportion of repetitive sequence (49.3%; Schlueter et al. 2008) and the proportion of methylated DNA (60%), obtained by methyl filtration of common bean (BAT93) DNA (P. Gepts, unpublished data). It is, however, larger than previous estimates for the heterochromatin content of the species (approximately 10% and 21%), which were based on centromeric DAPI⁺ bands after FISH (Moscone et al. 1999) and C-bands (Zheng et al. 1991), respectively. It is also considered high when compared to estimates for other species based on measurements of heterochromatin revealed by C-banding or CMA/DAPI staining (see, for example, Guerra 1993). Furthermore, it is similar to the estimated heterochromatin content from sorghum, which has a larger genome (818 Mb; Kim et al. 2005a), although smaller than estimates based on pachytene-condensed regions from M. truncatula (60% of 560–580 Mb; Kulikova et al. 2004). Interestingly, the pericentromeric heterochromatin of the species can be clearly subdivided into two different domains: the pericentromeric region, corresponding to BAC 12M3 distribution, CMA⁺ bands, probably the strongly condensed pachytene regions (Pedrosa-Harand, unpublished data), and the C-bands described in previous publications (Zheng et al. 1991; Moscone et al. 1999), and the centromeric region, revealed by DAPI after some FISH experiments and corresponding to only about 12% of the chromosome length. None of the hybridized BACs selected with genetically mapped markers showed a “centromeric” distribution, possibly due to the lack of single-copy dispersed sequenced in this domain. We hypothesize that the centromeric regions may be composed mainly of tandemly repeated sequences, while the pericentromeric regions surrounding the centromere are composed of dispersed repetitive sequences interspersed with single-copy sequences, leading to distinct stainability after different banding procedures of these two heterochromatic domains. Supporting this hypothesis, subcloning and sequencing of pericentromeric BACs identified retroelement-like sequences in their inserts (K.G.B. dos Santos et al., manuscript in preparation). This marked difference is, however, not clear in the centromere structure of other plant species studied so far, such as Arabidopsis, rice, and maize, because the centromeres of these species are composed of intermingled satellite repeats, centromere-specific retrotransposons, and genes (Nagaki et al. 2004; Ma et al. 2007; Yan and Jiang 2007; Gill et al. 2008). Therefore, it is possible that the proportion of these different types of sequences, rather than their presence or absence, defines these different chromatin domains.

It is worth noting that the bands obtained with DAPI after the FISH procedure are different from the CMA⁺ bands that, because of their extension, probably correspond to the previously reported C-bands (Zheng et al. 1991, 1993; Moscone et al. 1999). DAPI-after-FISH bands have been considered as largely equivalent to C-bands in common bean and other species (Moscone et al. 1996, 1999; Barros e Silva and Guerra 2009), but we demonstrate that they correspond to only a small fraction of the pericentromeric heterochromatin in common bean. Consequently, estimates of heterochromatin content based exclusively on DAPI-after-FISH bands should be considered with caution. A combination of different approaches is more likely to reveal the heterochromatin of a species in its totality and complexity.

We demonstrate that the genome of common bean is composed of 52% euchromatin enriched with single-copy sequences interspersed with dispersed repeats, 31% of pericentromeric heterochromatin (22%, excluding 12% of centromeric heterochromatin) and subtelomeric (9%) heterochromatin enriched with dispersed and tandem repeats interspersed with single-copy genetically mapped sequences, and 17% of rDNA (5%) and centromeric (12%) heterochromatin composed of repeats probably organized in tandem and largely devoid of low-copy-number genes. The tandem arrangement of the repeats in the subtelomeric heterochromatin of common bean has been demonstrated (David et al. 2009). This organization has implications for the strategy defined for sequencing the common bean genome. In practice, only the latter relatively small fraction (17%) could be left aside during a genome sequencing endeavor, considering that most of the heterochromatin also contains genes that should be sequenced. Therefore, a whole-genome approach, such as the selected whole-genome shotgun sequencing strategy (Scott Jackson, personal communication), is more appropriate. We should also anticipate the difficulty of dealing with a high proportion of repetitive sequences in the common bean genome (Schlueter et al. 2008) and possibly the need for further BAC FISH mapping for ordering contigs or estimating gaps, as performed in tomato (Peters et al. 2009). Furthermore, this organization suggests that the heterochromatin should be considered as a heterogeneous chromosomal domain (see Chang et al. 2008) even in a species with a small genome.