A hybrid BAC physical map of potato: a framework for sequencing a heterozygous genome

Abstract

Background: Potato is the world's third most important food crop, yet cultivar improvement and genomic research in general remain difficult because of the heterozygous and tetraploid nature of its genome. The development of physical map resources that can facilitate genomic analyses in potato has so far been very limited. Here we present the methods of construction and the general statistics of the first two genome-wide BAC physical maps of potato, which were made from the heterozygous diploid clone RH89-039-16 (RH).

Results: First, a gel electrophoresis-based physical map was made by AFLP fingerprinting of 64478 BAC clones, which were aligned into 4150 contigs with an estimated total length of 1361 Mb. Screening of BAC pools, followed by the KeyMaps in silico anchoring procedure, identified 1725 AFLP markers in the physical map, and 1252 BAC contigs were anchored the ultradense potato genetic map. A second, sequence-tag-based physical map was constructed from 65919 whole genome profiling (WGP) BAC fingerprints and these were aligned into 3601 BAC contigs spanning 1396 Mb. The 39733 BAC clones that overlap between both physical maps provided anchors to 1127 contigs in the WGP physical map, and reduced the number of contigs to around 2800 in each map separately. Both physical maps were 1.64 times longer than the 850 Mb potato genome. Genome heterozygosity and incomplete merging of BAC contigs are two factors that can explain this map inflation. The contig information of both physical maps was united in a single table that describes hybrid potato physical map.

Conclusions: The AFLP physical map has already been used by the Potato Genome Sequencing Consortium for sequencing 10% of the heterozygous genome of clone RH on a BAC-by-BAC basis. By layering a new WGP physical map on top of the AFLP physical map, a genetically anchored genome-wide framework of 322434 sequence tags has been created. This reference framework can be used for anchoring and ordering of genomic sequences of clone RH (and other potato genotypes), and opens the possibility to finish sequencing of the RH genome in a more efficient way via high throughput next generation approaches.

Figure 1

Example of a non-selective AFLP BAC fingerprint. (A) Original fluorescence trace file from BAC clone RH084E02. (B) AFLP band mobilities (in bp) and peak height values extracted from the trace file with band calling software. Only the band mobilities are used for fingerprint alignment by FPC.

Figure 2

Distribution of the number of bands per BAC in the 64478 fingerprints of the AFLP physical map. Counts of BACs in contigs and of singleton BACs are shown separately and are stacked to make the distribution of the complete set of fingerprints.

Figure 3

Flow diagram of the AFLP marker anchoring procedure of the potato AFLP physical map. The anchoring procedure is illustrated for genetic marker EAAGMCGA_326.6. (A) The marker from parent RH is traced back in the original radioactive gel of the genetic map. (B) As an intermediate step, a new radioactive gel is made with primer combination EAAGMCGA of the parental DNAs and 21 of the BAC superpools, and the position of the marker is identified. (C) The full set of BAC pools is examined with primer combination EAAGMCGA using capillary electrophoresis. The position of the marker is identified in these capillary fingerprints by comparison of the patterns of the first 21 lanes to those in (B). The band scoring interval of marker EAAGMCGA_326.6 was set to 324.0-324.3 bp and the average size was measured to be 324.1 bp. (D) Fingerprint bands are scored within this size interval and from the 31 positive superpool lanes (A02....H12) a list of quarter plate pool IDs is generated (001Q1... 192Q3) that are candidates for having the marker. (E) Seven of the QPP's identify the contig with the marker in the physical map, on the basis of matching BACs that have the marker band (e.g. BAC RH042F12 is from well F12 of library plate number 042, which is present in quarter plate pool 042Q4).

Figure 4

Genetic map locations of AFLP anchored BACs of the potato AFLP physical map. The genetic map of parent RH has twelve chromosomes (RH01 to RH012) and is made up from AFLP markers of the enzyme combinations EcoRI/MseI, SacI/MseI and PstI/MseI. Per chromosome, the genetic map is divided into up to 105 numbered bin segments that each represent a distance of one crossover event (0.77 cM) in the mapping population. The number of RH AFLP markers placed in each bin is indicated by a grey intensity value. Red bars indicate the counts per bin of BACs that are anchored to the genetic map by an EcoRI/MseI AFLP marker in their contig. For AFLP markers that mapped to a range of bins, the associated BAC counts have been evenly distributed over these bins. The bins with the centromere have their BAC count shown in blue and follow the identifications by Tang et al. [31] and Park et al. [42]. The BACs of the Nucleolar Organizer Region (NOR) do not have an AFLP anchor, but were identified by their end sequence. Chromosome orientations are according to bin number in the ultradense genetic map. For alignment to other potato and tomato genetic maps, e.g. from Tanksley et al. [43], chromosomes 7, 10 and 12 are in the wrong orientation and must be inverted.

Figure 5

Identification of the NOR in the AFLP physical map. Pachytene FISH of BAC clone RH127D02 showed its localisation in the compound structure of the Nucleolar Organizer Region (NOR) on the short arm of chromosome 2 (see Tang et al. [31] for methodology). Both brighter fluorescing regions and relatively weaker ones are visible, suggesting differences in NOR chromatin density. In the AFLP physical map the NOR is represented by a 96-clone BAC contig containing RH127D02. In the WGP physical map this NOR contig is absent.

Figure 6

Performance of the BAC superpool design in AFLP marker anchoring. (A) Numbers of placed (blue) versus unplaced (light blue) QPPs for markers with an increasing BAC copy number in the physical map. The resolved positive QPP counts show the innate capacity of the pooling design to accurately locate markers in the QPPs, which declines at high marker copy numbers. With the AFLP marker anchoring procedure, this decline in pooling design performance was compensated, and markers with relatively high copy numbers were identified on BAC clones without losses. (B) Distribution of the number of anchored BACs per AFLP marker. This figure closely represents the marker copy number distribution in the BAC pools.

Figure 7

Distribution of the number of WGP sequence tags per BAC for clones incorporated in the WGP physical map. (A) Distribution for 44292 clones from the RHPOTKEY library. (B) Distribution for 21627 clones from the RHPOTLUC library. Counts of BACs in contigs and of singleton BACs are shown separately and are stacked to make the distribution of the complete set of clones.

Figure 8

Frequency distribution of WGP sequence tags in 66545 BACs of potato clone RH. Sequence tags that were present in only one BAC clone were not included in the WGP dataset.

Figure 9

Theoretical frequency distribution of WGP sequence tags. This distribution is based on 8.2 genome equivalents of template DNA and 54% of heterozygous tags, and is a combination of two Poisson distributions for respectively the heterozygous and homozygous tags. It gives a good fit to the lower half of the observed distribution (Figure 8) and accommodates the relatively high fraction of two-copy tags in the WGP-dataset. However, other distributions with slightly higher g.e. and heterozygosity values will fit as well. The theoretical distribution assumes that all sequence tags are derived from single loci, and that no losses or errors have occurred with WGP sequencing. The relatively thick tail in the observed distribution (Figure 8) indicates that some of the tag sequences are likely to have come from duplicated loci.

Figure 10

Principle of the DNA superpool design of the RHPOTKEY BAC library. Pooled DNAs from 764 quarter library plates (quarter plate pools, QPPs) are each added to a unique combination of four different superpools (SP1...SP90), as shown here for three QPPs. Genetic marker screening is performed on the 90 superpools, and the marker-positive QPPs are then identified by deconvolution of the pooling design (see text for explanation).

Figure 11

Computer simulation of the behaviour of the BAC superpool design. Shown is the deconvolution output, in terms of QPP categories, for different input numbers of marker-positive QPPs in the pooling design. Resolved positive QPPs are input QPPs that are recognized with certainty by the pooling design as being positive. At high input numbers of positive QPPs, not all of them can be resolved as being positive anymore and, in addition, false positive QPPs begin to contaminate the output list.