Advances in high throughput sequencing technologies enable the mapping of genetic factors associated with multiple traits and the subsequent identification of candidate genes [35, 36]. Next generation sequencing, NGS, is widely utilized for QTL and GWAS (genome wide association studies) analyses to identify the regions responsible for a specific phenotype. Both QTL and GWAS methods have advantages and disadvantages. Biparental QTL analysis can identify a strong genetic factor but the confidence interval is affected by the number of recombination events in the population. GWAS utilizes a high number of recombination events within the population leading to the identification of narrower QTL intervals compared to biparental QTL mapping. However, association mapping studies are constrained because many QTL may be observed that explain only a small portion of the phenotypic variation. To overcome the limitations of QTL mapping and association mapping, and to narrow the interval of an important QTL for WM resistance in common bean, we used bulked-segregant analysis (BSA) coupled with marker selection [11]. To optimize the BSA, we used marker data from a preliminary biparental QTL analysis along with phenotypic data to select lines with resistant or susceptible haplotypic state to select lines for the DNA bulks. Two RIL populations which possess WM resistance from different genetic backgrounds were evaluated. Population R31 was developed from a cross of two black bean parents, while population Z0726–9 was derived from crossing pinto and great northern parents [9, 11] . QTL WM2.2 was previously discovered in the genotypes, I9365–31 and USPT-WM–1, the parents of the R31 and Z0726–9 populations respectively. Given the WM2.2 intervals were large in these populations, it was of interest to fine map this QTL.
In this study, the R31 QTL was detected within the same PV02 region that meta QTL WM2.2 was previously mapped [9] but with a narrower 3.57–4.54 Mbp interval. For Z0276–9, the QTL was identified in a broad region that overlaps the heterochromatin region of the genome. This broad region also overlaps QTL in the BV and R31 populations [9]. This large region may have resulted from the low level of recombination in the Pv02 heterochromatic region that is estimated to extend from 8.0–27.1 Mbp in the reference genome of common bean [36]. Fixed SNPs provide the greatest genotypic contrast between the two bulks. The significant peak interval in R31 was narrower than the Z0726–9 peak. This occurred because the R31 peak is located in the highly recombinogenic euchromatic region of Pv02 whereas the Z0726–9 peak is located in the low recombination heterochromatic region. This has been seen previously in common bean where the WM8.3 QTL was identified in the heterochromatic region of Pv08 across 10.75 Mb, while QTL WM7.1 mapped in the euchromatic region of Pv07 to a narrow 1.25 Mbp interval [11]. In one other comprehensive study of 771 QTL related to 161 unique sorghum traits, the heterochromatic region had a mean QTL density of 11QTL/0.5 Mbp compared to 7.5 QTL/0.5 Mbp in the euchromatic region [37]. Although QTL detected in the recombination-poor regions might complicate successful gene cloning, they can be very useful in MAS breeding programs. Moreover, the availability of a P. vulgaris reference genome sequence will help overcome the complication of gene cloning in these regions.
Combining the approaches of QTL with a greater number of markers and sequence-based introgression, it was determined that the meta-QTL WM2.2 consists of two independent QTL regions WM2.2a ( = 1.67 Mbp) and WM2.2b ( = 14.22 Mbp). WM2.2a, physically mapped in the euchromatic region, and WM2.2b in the heterochromatic region, may provide different WM tolerance/resistance mechanisms.
Using only significant fixed polymorphism sites between the two extreme groups of phenotypes allowed us to select candidate genes that were highly polymorphic between susceptible and resistance parent. Furthermore, using shared sites that differed by allele frequency of 0.3 or more, a minor QTL was detected within a small interval that corresponded with meta-QTL WM5.4 in the R31 population. This suggests that WM2.2 and WM5.4 may interact with each other in the R31 population to provide a higher level of resistance [9]. Previously in the fine mapping of the WM7.1 QTL, the resistant allele for QTL WM9.1 was also detected, and it was hypothesized that the interaction of the two loci provided a higher level of resistance to WM disease [11].
Our fine mapping revealed potential candidate genes associated with synonymous and/or non-synonymous fixed sites in WM2.2a and WM2.2b intervals. Pentatricopeptide repeat (PPR) proteins and gibberellin 2-oxidase are two important disease tolerance/resistance associated gene models detected in WM2.2a. PPR proteins share sequence features with NLR disease resistance genes particularly for their involvement in diversifying selection process. It has been shown that the birth and death process described for immunoglobulin genes [38], and later for plant R genes [39] is also the route for tandem repeat duplications in the PPR protein gene family [40]. Many studies have shown that PPR proteins contain many sequence-specific RNA binding sites involved in several RNA processing activities [41, 42, 43, 44, 45]. For example, the mRNA encoding PPR proteins are the target for miR400 that down regulates the PPR genes by cleaving their mRNAs and subsequently the plant becomes more susceptible to pathogenetic bacteria and fungi [46]. Moreover, recent studies on sequenced genomes of several plants revealed that the nucleotide binding site of PPR protein genes are similar to the NB-ARC nucleotide binding site of NLR genes which are part defense mechanism used by plants to respond to pathogen infection [47].
Gibberellin 2-oxidase is candidate gene that may explain WM disease avoidance mechanism associated with plant architecture. Gibberellins controls inter-node length and therefore indirectly the shape of the bean plant canopy. Shorter internode plants have an upright canopy that allows airflow through the field which in turns reduces the high humidity environment favored by the white mold pathogen. Long internode plants become prostrate on the ground which leads to a low, dense canopy with high humidity conditions favored by the pathogen. This leads to an increased severity of WM disease in the field [5, 48]. Gibberellin 2-oxidase inactivates gibberellins and effectively changes the status of this important hormone for internode extension. This feature makes gibberellin 2-oxidase an important candidate gene because WM2.2 is associated with canopy porosity, canopy height, and lodging in the R31 population. Vasconcellos et al. [9], detected a QTL for canopy porosity, co-located (5.8 Mb) with WM2.2 in R31 population. Furthermore, WM2.2 was not detected in the greenhouse straw test used to detect physiological resistance. This further supports avoidance related genes as primary candidates. Moreover, significant fixed SNP sites detected in the WM2.2a interval were located inside the start site of the gibberellin 2-oxidase gene which makes it a reasonable candidate gene for disease avoidance traits.
Hsp70, a candidate gene in the WM2.2a interval, interacts with other plant defense genes [49]. HSPs are molecular chaperons proteins which play an important role in modulating the structure of disease resistance proteins, and are found to modulate Arabidopsis defense against pathogens [50]. In another study, Jelenska et al. [51], reported that the P.syringae pathogen uses the HopI1 effector to hijack Hsp70-assosiated chaperoning activity. Under heat stress conditions, if excess HSP70 is provided, this effector is dispensable for P.syringae pathogenicity. This suggests searching for candidate genes associated with disease resistance, in addition to looking for NB-ARC genes in significant intervals, other genes which might be the target of a pathogen effector should also be consider. These genes/proteins might or might not have a R gene domain structure, but rather may be critical cellular proteins that directly or indirectly influence R-gene functions by modulating their structure or/and stability following their interaction with an effector [52].
The significant WM2.2b interval discovered in the Z0726–9 population was identified in the heterochromatic region. A cluster of genes encoding LRR proteins, typical of plant disease associated resistance genes, were detected in this region. This finding suggests a role in physiological resistance for WM2.2 in the Z0726–9 population. Partial physiological resistance is important when the severity of the disease overcomes the avoidance mechanisms provided by upright architecture or, reduced lodging. However, when all of the LRR clusters in this region (19.78–20.21 Mbp) were analyzed in Pfam [53], only a single TIR-NBS-LRR gene model (Phvul.002G079200) possessed all NB-ARC domain amino acid signatures necessary for a functional disease resistance protein. Among the gene models associated with non-missense variants in the WM2.2b interval, an EFR gene (Phvul.002G124300) is an important part of plant immune response. This receptor is known as a pattern-recognition receptor (PRR) with an extracellular domain harboring the LRR domain which recognizes and binds to the prokaryotic elongation factor EF-Tu [54]. This response prevents subsequent spread of the infection through the plant after the plant becomes infected by the pathogen. However, this receptor can only be effective if the pathogen expresses an excess level of EF-Tu [55].