A backcross population was developed by crossing the susceptible diploid S. cardiophyllum (PI186548) as the male parent with the resistant diploid S. pinnatisectum (PI275233). Several clones of a single resistant F1 individual, propagated through in vitro culture, were then backcrossed with the susceptible parent to generate a backcross mapping population.

The F₁ and BC₁ populations were maintained vegetatively from tubers following their first propagation from true seed. A total of 931 clones from the backcross population were selected to analyze the genetics of resistance to late blight using detached leaf methods at the first clonal generation (Chen et al., 2003).

An inoculum was prepared from the P1801C.16 strain of P. infestans (US-8/A2 mating type) and diluted to a final concentration of 30,000 sporangia per ml. Inoculation and the detached leaf assay were performed according to Chen et al. (2003). Three compound leaves were excised for the late blight test, including the third to fifth leaves from the top on each plant’s main branch. Each compound leaf with five leaflets was inserted into prepared moist vermiculite in a plastic tray. The inoculum was sprayed onto the surface of all leaflets. Trays with inoculated compound leaves were incubated in a growth cabinet under an 18/6 h and 20/18°C day/night regime for about 15 days. The susceptible parent was inoculated as a susceptible check. Plant resistance was evaluated after 8 and 15 days. Disease severity was estimated using mean disease severity values (DSVs) of three compound leaves based on the percentage of leaf area with symptoms of late blight. Severity values were scored using a scale of 0–5 where 0 = no disease to <3%; 1 = 3–24%; 2 = 25–49%; 3 = 50–74%; 4 = 75–94%; and 5 = 95–100% infection. Plants with a DSV of 0 were classified as resistant and those with DSVs of 2–5 were classified as susceptible.

DNA was extracted from 100 mg of young leaves for each potato plant using a Genomic DNA Purification Kit (Promega, Fitchburg, WI, United States). Bulked segregant analysis (BSA) was used to screen for molecular markers associated with late blight resistance (Michelmore et al., 1991). Two susceptible bulks were constructed from 10 highly susceptible individuals and a resistant bulk was developed with equal amounts of DNA from 10 highly resistant individuals among the BC₁ population. DNA markers were screened for the two susceptible bulks, the resistant bulk and the resistant parent plant.

AFLP analysis was performed as described by Vos et al. (1995) using EcoRI and MseI as rare- and frequent-cutter enzymes, respectively. Genomic DNA digestion and ligation were conducted using an AFLP Core Reagent Kit (Invitrogen, Carlsbad, CA, United States) according to the instructions. A pre-amplification was carried out with 1-bp extension primer combinations (EcoRI+A/MseI+C and EcoRI+A/MseI+A) and the PCR products were diluted at a ratio of 1:30 with TE buffer. Selective amplification using primer combinations of EcoRI+3 and MseI+3 was conducted and the products were separated on a 6% PAGE sequencing gel run at 100 W for 2.5 h after pre-electrophoresis for 30 min. The gel was stained by the silver-staining method (Bassam et al., 1991).

AFLP fragments were excised from the dried silver-stained polyacrylamide gel and placed into microfuge tubes containing 30 μl distilled water. The samples were boiled for 10 min and centrifuged, and then 3 μl of the supernatant was used for PCR under the same conditions as those used for AFLP analysis. The PCR products were then inserted into the pGEM-T easy vector (Promega, Fitchburg, WI, United States) and sequenced. A search for sequences homologous to the AFLP fragments was conducted using the GenBank website¹, and Clustal W2² was used to compare the sequence homology.

The digestion of genome DNA were performed by two restriction enzymes, EcoRI and MseI, according to the instructions of AFLP Core Reagent Kit. The primer combination EcoRI/MseI was used to generate pre-amplification products. Then, the second amplification step was carried out with five primer combinations. The primer combinations were respectively combined by five P-loop based RGAs primers, S1 (5′-GGTGGGGTTGGGAAGACAACG-3′), S2 (5′-GGIGGIGTIGGIAAIACIAC-3′), AS1 (5′-CAACGCTAGTGGCAATCC-3′), AS2 (5′-IAAIGCIAGIGGIAAICC-3′) and AS3 (5′-IAGIGCIAGIGGIAGICC-3′) (Leister et al., 1996), with the EcoRI AFLP primer. PCR conditions were somewhat different from the standard AFLP procedure; 30 s DNA denaturation at 94°C, 30 s primer annealing at 55°C and a 1 min elongation step at 72°C (35 cycles). Prior to the cycling, the template DNA was denatured for 1 min at 94°C and the PCR was finalized by applying an extra 5 min elongation step at 72°C. The procedures for running the gel and fragment extraction were the same as described for AFLP section.

Locus-specific markers on chromosome 7 of potato and tomato³ were selected to develop PCR-based markers. Generally, the RFLP probe sequences were used as queries to search ESTs using the BLASTn program (Altschul et al., 1997). Then, the ESTs and RFLP probes were assembled with a criterion of more than 95% identity over a stretch of 40 nucleotides using SeqMan II (LASERGENE, DNASTAR, Madison, WI, United States). Primers were designed according to the assembled sequences guided by intron finder⁴ to amplify regions spanning introns and avoid placing primers in exon–intron boundaries. The PCR products were separated after digestion with one of the 4-bp cutter restriction enzymes TaqI, Trul1, Msp1, Rsa1, and Tail1.

The PCR amplification reactions were conducted in 20 μl reaction mixtures containing 10 mM TRIS-HCl, pH 8.3, 50 mM KCl, 2 mM MgCl₂, 100 μM of each dNTP, 200 nM primers, approximately 20 ng template DNA and 1 Unit Taq DNA polymerase. The cycling program consisted of an initial 3 min denaturation step at 94°C, followed by 35 cycles of 94°C (30 s), 55°C (30 s), and 72°C (50 s), and a final 5 min extension step at 72°C. The PCR products were size-separated on a 2% agarose gel, stained with ethidium bromide, and visualized on a Gel Imaging system (Bio-Rad, San Diego, CA, United States).

Linkage analysis was performed using the software package MAPMAKER V3.0 (Lander et al., 1987). Markers and their corresponding distances (cM) were included in the framework map only if the LOD value for the ripple was >3. The Kosambi mapping function was employed to convert recombination frequencies to map distances in cM (Kosambi, 1943). Collinearity analysis results were visualized using Circos-0.67 (Krzywinski et al., 2009).

In an attempt to find AFLP markers linked to the resistance locus, 324 EcoRI+3/MseI+3 (196 E-A/M-C and 128 E-A/M-A) AFLP primer combinations were screened in the bulk material using a BSA strategy. Ten putative AFLP fragments were identified and segregation analysis in the BC₁ population confirmed that seven of them were associated with the resistance locus. The two closest markers, EAGCMCGA-450 and EACAMAGG-330, were determined to be linked to the resistance locus at distances of 1.2 and 0.8 cM, respectively.

These two AFLP fragments were cloned and sequenced. BLAST analysis showed that the sequence of EAACMATC-330 had no similarity to any known sequence in GenBank, whereas EAGCMCGA-450 hit four potato ESTs (BQ509088, BG600948, DV623421, DV623416). These four ESTs and two other potato ESTs were assembled into a 1720-bp contig with a complete coding region that showed high similarity to the Arabidopsis gene GLUCAN SYNTHASE-like 7 (1e-139) in Blastx⁵) analysis. Based on this assembled sequence, a CAPS marker of EAGCMCGA-450, SpAFLP1, was developed (Figures 1A,D). In addition, EAACMATC-330 was converted into the CAPS marker SpAFLP2 (Figures 1B,E).

The 1720-bp contig of EAGCMCGA-450 was used to search the high-throughput genomic sequence (HTGS) database with BLASTn⁶ and a tomato BAC (C07HBa0116M01) was identified (e-112). Annotation of this BAC (C07HBa0116M01) revealed a partial VPS16-like gene in the 3′ terminus. This partial gene sequence was used as a query to search the EST database with BLASTn and 23 matching ESTs were identified. All 23 ESTs were assembled to a 2.4-kb contig, named cEST1. BLAST analysis showed that this sequence had homology to an RFLP marker, TG572 (e-120), which was mapped to tomato chromosome 7. Subsequently, a CAPS maker named SpTG572 was developed according to this sequence and was shown to co-segregate with SpAFLP1 (Figure 2).

The 5′ terminal sequence of BAC C07HBa0116M01 was used to build a 2-kb contig with 10 ESTs, named cEST2. This contig hit a potato BAC end sequence, POTDQ81TR. Using the other end sequence POTDQ81TF as a query, we identified a tomato BAC, C07HBa0018L21. A PCR marker, SpAL21, was developed based on the left end sequence of BAC C07HBa0018L21 and recombination was found between this marker and SpTG572 (Figure 2).

TG572 was near I-3, a gene for fusarium wilt resistance from the wild tomato species Lycopersicon pennellii, with a genetic distance less than 0.3 cM (Hemming et al., 2004). Two additional markers closely linked to I-3, CT226, and Got2, were converted to SCAR markers in our mapping population, and named SpCT226 and SpGot2, respectively. Segregation analysis indicated SpCT226 and SpGot2 were proximal and distal with genetic distances of 2.8 and 3.2 cM, respectively (Figure 3B).

The flanking markers TG572, T0810, and T1756 in the SGN map⁷ were developed into PCR-based markers and tested in the mapping population. The results delimited the resistance locus to the interval between StAFLP2 and SpT1756 on potato chromosome 7 (Figure 3B).

The RGA fingerprinting technique was used to identify functionally relevant markers linked to the resistance for late blight. An RGA fragment, RGA1, amplified by the AS2 primer (Leister et al., 1996) in combination with the E00 AFLP primer, did not exist in the two susceptible bulks but appeared in the resistant bulk and the resistant parent (Figure 1C). Sequence analysis of this 320-bp long fragment revealed homology to an RGA sequence previously mapped to the long arm of potato chromosome 7, Gro1-5 (Leister et al., 1996). Therefore, this fragment was named SpGrol-1. A PCR marker was developed and segregation analysis indicated that SpGrol-1 was 2.8 cM from the resistance locus (Figure 1F).

Previously, the late blight resistant locus Rpi1, also derived from S. pinnatisectum, was assigned to chromosome 7 flanked by two RFLP markers, TG20 and CP56 (Kuhl et al., 2001). To compare the map positions of Rpi1 and our target resistance locus, the RFLP markers TG20, CP56 and their interval marker GP127⁸ were converted into PCR-based markers. The marker information is listed in Supplementary Table S1. Segregation analysis of these converted PCR-based markers showed a link between the late blight resistance loci. The genetic distance between CP56 and TG20 was 15.1 cM, similar to the map of Kuhl et al. (2001) (14.6 cM). However, there was an obvious difference in the genetic distance between the markers linked to the resistance genes. CP56 and TG20 were 4.0 and 11.1 cM away from our target gene, respectively, which was different to the genetic distances of Rpi1 to these two markers (9.4 and 5.2 cM, respectively) (Figures 3A,B). Therefore, our resistance gene was called Rpi2.

Although the two resistance genes were derived from the same species, they came from two different accessions; Rpi1 from PI235214 and Rpi2 from PI 275233. To further identify genetic differences between the accessions PI275233 and PI235214, we screened all of the markers and found that the two accessions had different genotypes at two loci, SpAFLP2 and SpCT226.

The molecular marker sequences were used as queries to search for homologous loci in the genome sequence databases of potato and tomato. The majority of markers linked to Rpi2 showed homology to STChr7 and SLChr7, and were included in two orthologous genomic regions spanning 2.9 and 2.4 Mb, respectively (Figure 4B). Specifically, 9 of the 11 molecular markers mapped in the Rpi2 genetic linkage map generated hits to 16 homologous loci in STChr7 and 10 loci in SLChr7 (Figures 3C,F). Because cloning and sequencing failed, homologous loci of SpGot2-1 and SpTG20 were not found.

Comparative genomic analysis revealed that 14 of the 17 annotated genes in potato between StAFLP1 and StAL21 had similarity to corresponding regions in tomato (Figures 3D,E), again revealing high levels of collinearity in the Rpi2 region between potato and tomato. Furthermore, the order of these similar genes was highly conserved but reversed. Segmental inversion, which was a reasonable explanation for the reverse order of the conserved genes, was observed on STChr7 between StAFLP1 and StAL21 compared with the Rpi2 linkage map and SLChr7.

Scanning the Spud DB Genome Browser for Potato (Solanum tuberosum group Phureja DM1-3) PGSC v4.03 Pseudomolecules⁹ suggested the physical distance between StAFLP2 and StT1756 was about 84 kbp. Consequently, we estimated that the physical distance between Rpi2 and SpAFLP2 was about 28 kbp by referring to their genetic distance. Unfortunately, there was a 50-kbp gap in this potato genome region that potentially contained the homologous gene of Rpi2 (Figure 4C). However, the homologous segment in the tomato genome was assembled completely, in which 3 of 10 loci Table 1 were annotated as NBS-LRR class disease resistance proteins (Accession nos. Solyc07g056180.1, Solyc07g056190.2, and Solyc07g056200.2) (Figure 4A).

Accumulated evidence has suggested that resistance loci are not distributed randomly along chromosomes. Several hotspots for resistance genes have been described in Solanum species. For instance, at least five R genes against diverse pathogens have been mapped to the GP21–GP179 interval on chromosome 5 in different genetic backgrounds; Gpa and Grp1 conferring resistance to potato cyst nematodes (Kreike et al., 1994; van der Voort et al., 1998), Nb and Rx2 conferring resistance to potato virus X (Ritter et al., 1991; DeJong et al., 1997), and R1 conferring resistance to P. infestans (Leonards-Schippers et al., 1992). In the current study, we found that Rpi2 was located in a major cluster on the long arm of chromosome 7 in which several R genes have been mapped, including Rpi1 conferring resistance to P. infestans, Gro1-4 conferring resistance to Globodera rostochiensis, and I-3 conferring resistance to Fusarium oxysporum (Bournival et al., 1989; Ballvora et al., 1995; Kuhl et al., 2001; Paal et al., 2004; Ruggieri et al., 2014; Catanzariti et al., 2015). Clearly, this region is another hotspot for resistance genes, and can be expected to contain more resistance genes.

Resistance loci regions are usually enriched in NBS-LRR homologs. For instance, there are at least 13 TIR-NBS-LRR sequences clustered across more than 400 kb in the locus Gro1 (Paal et al., 2004) and one of them, Gro1-4, has been shown to be responsible for a resistance trait. In this research, we used the RGA profiling strategy to identify an RGA fragment, SpGrol-1, linked to the resistance locus. Sequence analysis showed that SpGrol-1 belonged to the TIR-NBS-LRR family, and that the most similar sequence was Gro1-5, a gene at the Gro1 locus. However, mapping analysis showed this RGA was proximal to Rpi2 with a genetic distance of 2.8 cM.

Resistance gene analogs are generally clustered in the genome (Meyers et al., 1998; Michelmore and Meyers, 1998). Clusters of R genes can be tightly organized or spaced over several megabases of sequence (Meyers et al., 1998; Noël et al., 1999). We thought that similar Gro1-like sequences might be present in our resistance locus. Hence, we designed a set of primers according to an alignment of 13 Gro1 sequences and developed three PCR markers. However, all of these markers co-localized to SpGrol-1 (data not shown). This indicated that there was more than one Gro1-like sequence and that the Rpi2 gene might not be a Gro1-like gene.

A similar observation was also described for the resistance gene I-3 from the wild tomato L. pennellii (Hemming et al., 2004). I-3 co-segregated with RGA St332; however, RGA St332 was ruled out as a candidate gene for I-3 because it was a single-copy pseudo gene in L. pennellii. I-3 was flanked by two RFLP markers TG572 and CT226 in an interval of 0.6 cM. That Rpi2 and I-3 share the flanking markers CT226 and TG572 supports that these two genes are in a syntenic region.

Comparative genomics between potato and tomato facilitated the mapping and isolation of the late-blight R genes R3a and Rpi-blb2 from potato in a previous study, as these genes were mapped to regions of the potato genome that were syntenic to previously cloned gene loci (I2 and Mi, respectively) in tomato (Huang et al., 2005; van der Vossen et al., 2005). Recently, both the potato and tomato genomes have been sequenced (Xu et al., 2011; Sato et al., 2012). This sequence information should greatly accelerate the cloning of the Rpi2 gene through comparative genomics.

Comparing the homologous regions in the potato and tomato genomes, the genetic linkage map of Rpi2 showed high uniformity except that a chromosome inversion had occurred in the sequenced DM genome (Figure 3). Although this inversion may be a result of chromosomal variation during evolution, incorrect sequencing or assembly could equally have led to the observed recombination because short reads, a large amount of repetitive sequence, the sequence GC composition and other effects can impede uniform and complete sequencing coverage along the genome (Maiti and Bouvagnet, 2001; van Hijum et al., 2005; Aird et al., 2011; Schatz et al., 2012; Berlin et al., 2015). The gap between SpAFLP2 and SpT1756 on StChr7 indicated the accuracy of the assembly in the Rpi2-related region was not sufficiently high. In other words, the fragment inversion observed by comparative analysis was not sufficient evidence to demonstrate chromosome inversion. Furthermore, the lack of sequence information between the flanking markers prevented us from obtaining candidate genes from the DM genome data. Therefore, constructing a higher quality genome assembly for the Rpi2-related region requires enhanced approaches.

At present, an effort to introgress disease resistance genes from S. pinnatisectum into potato is being carried out to develop resistant cultivars. Because of the ploidy level barrier and endosperm balance number incompatibility, it is difficult to transfer resistance traits from S. pinnatisectum to cultivated potato. Fortunately, the hybridization barrier between S. pinnatisectum and cultivated potatoes can be overcome by embryo rescue, protoplast fusion, and chromosome-doubling techniques (Chen et al., 2008; Sanetomo et al., 2014). Our molecular markers could help breeders to introduce this resistance gene into different cultivars by marker-assisted selection.