Linkage disequilibrium at the APA insecticidal seed protein locus of common bean (Phaseolus vulgaris L.)

Plant Material

DNA extraction

DNA extraction technique was a standard CTAB(2%) and organic-solvent (chloroform-octanol) extraction based on the method of Afanador et al. [37] where 200 μl of extraction buffer (150 mM Tris-HCl, 15 mM EDTA, 1 M NaCl, 1.5% CTAB, 2% β-Mercaptoethanol) was added to 0.5 g of young trifoliate leaf tissue that was ground with a plastic pestle. A further 600 μl of extraction buffer was added and the mixture incubated at 65°C for 45 minutes before adding chloroform:octanol (24:1) mix, shaking for 30 minutes and precipitating the supernatant with 500 μl of isopropanol at -20°C and 125 μl of sodium acetate in a new eppendorf tube. The DNA was pelleted by centrifuging at 12,000 rpm for 10 minutes and cleaned with 500 μl of 70% ethanol. The DNA pellets were then dried and re-suspended in 150 μl of ddH2O to a concentration of 50-100 ng/μl. DNA was then diluted to 10 ng/μl for use as template for the amplification of microsatellites.

Marker amplification

This study used diversity evaluations for 100 microsatellites, five of which were selected for characterizing the APA region while the rest were based on the results of Blair et al. [34] and were from map locations throughout the genome. PCR conditions for the markers consisted of initial denaturation for 94°C for 5 minutes followed by 30 cycles of 94°C for 1 minutes, 47°C for 1 minute and 72°C for 2 minutes, ending with a final extension period of 72°C for 5 minutes. The microsatellites were run in 4% polyacrilamide (29:1 acrylamide:bisacrylamide) gels at 150 constant watts (1800-2000 volts) and 45°C constant temperature in Biorad Sequi-Gen GT vertical gel rigs for 1.5 hrs. Gels were silver-stained with a re-circulating tank system as described in Blair et al. [36]. Band sizes were determined by comparison to a molecular weight standard based on 10 and 25 bp ladders (Invitrogen, Carlsbad, CA).

Bruchid resistance screening

Techniques for insect rearing and testing of genotypes for bruchid resistance were similar to those described by Schoonhoven et al. [38]: bruchid resistance was evaluated with 3 repetitions of 30 seeds infested with 6 pairs of Z. subfasciatus (Boheman) in a mesh-covered, clear plastic vial (9 cm high × 1.7 cm in diameter) whose walls were covered with sandpaper (No. 150, rough side of sandpaper facing inwards) to avoid egg-laying on the plastic surface rather than the bean seed coat. Experiments were conducted in a rearing chamber at 27°C and 70% RH. Data was collected on number of eggs at 15 days after infestation at which point initial insect parents were removed and on the number of adults emerging before 50 or up to 70 days after infestation. The percentage emergence was calculated based on the total number of adults emerged compared to the number of eggs laid. Genotypes with 0 to 15% total adult emergence were classified as highly resistant (HR), from 15 to 30% as resistant (R), from 30 to 50% as intermediate (I) and from 50 to 100% as susceptible (S).

Protein extraction and arcelin determination

The presence of the Arcelin protein and the type of arcelin variant were determined according to the methods described by Ma and Bliss [39] where 0.75 g of bean seed flour was dissolved in 250 ul of extraction buffer (1 M NaCl, pH 2.4), vortexed and centrifuged at 14,000 rpm for 15 minutes, the supernatant was transferred and mixed with 50 μl of cracking buffer (100 mM Tris-HCl, 1% SDS, 20% sucrose, 0.05% β-Mercaptoetanol, 0.005% Bromphenol blue) before being vortexed again, boiled for 5 minutes, allowed to cool and centrifuged again for 15 minutes at 14,000 rpm. A 5 μl aliquot of this mixture was loaded onto a stacking 4% polyacrylamide gel in a Biorad Mini Protean 3 Cell gel apparatus that was run at a constant 150 volts until the sample passed into the running gel at which time 25 mA current and 20°C temperature were maintained. Protein gels were stained for 4 to 5 hours in 120 ml of 0.025% Coomassie Blue R-250, 45.4% methanol, 9.2% acetic acid, and 45.4% distilled water then transferred to distaining solutions (I: 10% acetic acid, 50% methanol and II: 7% acetic acid, 50% methanol) for 4 to 5 hours.

Linkage disequilibrium and association analysis

LD analysis was carried out for the arcelin resistant and susceptible germplasm in Table 1 for the markers described above. LD values for pairs of markers around the APA locus were estimated from the allele data using the software program PowerMarker v. 3.0 [40] with LD measured as the square of the correlation coefficient (r²) between alleles at the two loci. LD matrices were visualized by 2-D plot in the same program. LD was calculated for the complete dataset as well as for the sets of resistant wild and cultivated genotypes separately. Given population structure for the cultivated genotypes, these genotypes were separated into Andean and Mesoamerican gene pools for additional evaluation of LD with the same procedure. All LD values were tested for significance with Fisher's exact tests and the P-values were reported. Associations between bruchid resistance traits and genetic markers near or in the APA locus were determined for the genotypes based on a fixed effect linear model in Trait Analysis by aSSociation, Evolution and Linkage (TASSEL) software http://www.maizegenetics.net/tassel. LD estimates were also made for the principal subpopulations for the 105 common bean genotypes from the cultivated Andean and Mesoamerican genepools and from the wild Mesoamerican sources of resistance to bruchids, These subpopulations were confirmed with STRUCTURE [41] analysis was conducted using K = 2 and K = 3 with 50,000 burn-ins and 100,000 iterations using an admixture model in each case.

Germplasm survey showed arcelin specific microsatellite alleles

To begin with, this study focused on some previously developed arcelin region markers selected because they targeted sequences of the APA gene family (BMd9, BMd15, BMd16, PV-ag004 and PV-atct001) as shown in Figure 1 or because of their map location and proximity to the arcelin locus (BMd26, BMd8 and BMd30) on linkage group B4 [34, 35]. Among the APA based markers, BMd9 was derived from the gene for D-Lec 2 Phytohemagglutinin-L (Genbank entry X06336), BMd15 from Erythroglutinating Phytohemagglutinin (K03288), BMd16 from Leucoaglutinating Phytohemagglutinin (K03289), PV-ag004 from a Phytohemagglutinin pseudogene (X04660) and PV-atct001 from upstream of the arcelin gene (M68913). BMd9 and BMd16 were found to produce background bands, however all other markers worked well and were evaluated for allelic diversity.

Therefore, with the goal of evaluating the LD around the APA locus and association of this locus with bruchid resistance, the selected microsatellite markers were evaluated in arcelin-containing resistant genotypes including the wild bean sources of the gene, advanced breeding lines with the designations RAZ and GG as well as a corresponding set of bruchid-susceptible genotypes that have been used in breeding programs targeting resistance to Z. subfasciatus. The alleles for the five polymorphic and reliable markers are shown in Table 1 with BMd30 as monomorphic and therefore not discussed further.

Microsatellite allele diversity was found to vary with each of the markers tested in the germplasm evaluation. The marker detecting the most alleles was PV-ag004 with 10, while PV-atct001 detected 5 alleles, BMd26 detected 4 alleles, and BMd8 and BMd15 detected 3 alleles each. Within each of the subgroups of germplasm, the diversity observed with the different microsatellites also varied. For example, in the wild accessions that were sources of resistance, BMd15 detected only 1 allele, BMd26 detected 3 alleles, PV-atct001 detected 5 alleles, and PV-ag004 detected 7 alleles. Different banding patterns in the wild accessions were interpreted as separate alleles in the case of PV-ag004 and PV-atct001.

Among the markers, BMd15 was notable in that the allele in the resistant genotypes was different from the alleles in the susceptible genotypes and therefore was predicted to make a good diagnostic marker for bruchid resistance introgressions. The marker PV-ag004 was also diagnostic for the different arcelin alleles and resistant genotypes derived from these despite producing a multiple banding pattern. For example, all the Arc5 containing genotypes except for one had the 184, 203 and 207 bp bands, while all the Arc1 containing genotypes had the 184, 195 and 207 bands and therefore these could generally be distinguished. Meanwhile, all the susceptible genotypes had either of two patterns one with 203 and 242 bp bands and the other with 203 and 275 bp bands. For the PV-atct001 marker Arc5 containing lines could not distinguish from susceptible genotypes however Arc1 containing lines could.

Distribution of bruchid resistance and association with arcelin linked markers

The results of screening for insect resistance in the germplasm survey are shown in Table 1 and were correlated with the arcelin alleles detected. Among the wild beans, Arc1, Arc4, Arc5 and Arc7 containing accessions had higher resistance (HR) than Arc2, Arc3 and Arc6 containing accessions, which were intermediate (I) or susceptible (S). These results are consistent with those obtained in previous studies by Cardona et al. [29]. All the GG and RAZ series lines with Arc5 or Arc1 alleles respectively were found to be resistant or highly resistant, while all the genotypes that did not contain arcelin alleles both from the Mesoamerican and Andean genepools were found to be susceptible. Interestingly, the Arc1 allele generally provided higher resistance than the Arc5 allele in the cultivated background based on the comparison of the GG series lines versus the RAZ series lines. However this difference was not seen in the wild genotypes with these alleles since G12882 and G02771 had the same levels of bruchid resistance.

The wild beans were more variable than the resistant breeding lines perhaps given the multiple alleles evaluated in the wild (7) versus the alleles incorporated into RAZ or GG lines (2). More adults emerged at 70 days for the resistant genotypes than for the susceptible genotypes due to the slow development of the insects in the resistant lines. In other words, for the susceptible genotypes the seeds were already completely perforated and consumed so no more adults emerged in the period from 50 to 70 days, while in the resistant genotypes few adults emerged either at 50 or 70 days. As a result percentage emergence was highly contrasting between wild and cultivated resistant genotypes (16.4 and 5.0%, respectively) versus the susceptible genotypes (93.8%).

The higher percent emergence among the wild resistant genotypes can be explained by the variability in the different arcelin/APA alleles which led to some variability for resistance with some genotypes being intermediate (Arc2) or even susceptible (Arc3 and Arc6) in our assay although Arc1, Arc4, Arc5 and Arc7 were highly resistant. One interesting point is that fewer eggs were laid on the wild beans (48.0) than on the cultivated beans (94.0 on average). This is likely to have occurred because the wild beans are smaller seeded and therefore can host fewer eggs than the large seeded cultivated beans. Figure 2 shows the relative size of the wild beans containing the Arc1 and Arc5 alleles and two RAZ lines with arcelin-based, bruchid resistance introgressed into a cultivated red mottled background. It is notable that egg laying is similar on the cultivated resistant and cultivated susceptible genotypes since arcelin does not affect antibiosis at the level of oviposition but rather during larval development [2].

Linkage Disequilibrium near the APA locus and throughout the bean genome

Part of the observed LD would be explained by population structure since we included both Andean and Mesoamerican beans in the overall germplasm survey and because we included genotypes bred intentionally for bruchid resistance as well as susceptible genotypes. Therefore, we subdivided the survey into sub-populations based on STRUCTURE analysis finding separation of groups based on gene pool identity at K = 2 and the wild beans diverging at K = 3 as shown in Figure 3. This was significant since the Andean and Mesoamerican genepools are the major source of population structure in common beans but wild beans are very diverse [31–33].

By doing this we found that LD at the APA locus was lower in Andean genotypes than in the Mesoamerican genotypes. Meanwhile LD values between the two APA associated markers, PV-atct001 and PV-ag004, remained high in the cultivated resistant and wild resistant genotypes with r² of 0.9163 and 0.5553, respectively. Indeed, while the allele for BMd15 found in the wild sources was unique to the wild accessions and the derived introgression lines (RAZ or GG designation) and not present in any of the susceptible lines; the alleles of the markers BMd26 and BMd8, which map further away from the arcelin locus according to Blair et al. [34], presented a much lower association with the arcelin introgression or LD with the APA locus markers PV-ag004 and PV-atct001 in the cultivated resistant lines. For example, LD values between the APA locus markers and the markers BMd26 and BMd8 ranged from r² of 0.1172 to 0.2006 for cultivated resistant genotypes and r² of 0.0202 to 0.0990 for cultivated susceptible genotypes.

LD values with APA locus markers were compared with those from around the common bean genome and are shown in Figure 4. LD values were averaged per linkage group and ranged from r² of 0.1425 (for B3) to 0.4512 (B9) when considering only Mesoamerican genotypes and from r² of 0.2720 (B8) to 0.6590 (B9) when considering only Andean genotypes. Apart from B9, high average LD was also observed for Mesoamerican genotypes on linkage groups B1, B6 and B7 (all with average r² above 0.3) and for Andean genotypes on B6, B7 and B11 (all with average r² above 0.45). It was notable that LD values across the genome were slightly higher when considering the Andean genotypes than when considering the Mesoamerican genotypes.

The germplasm survey revealed varying levels of marker diversity

As part of the process to evaluate the microsatellite markers associated with APA genes we carried out a germplasm survey to determine the allelic diversity for the markers. In this part of the study, we found the number of alleles and level of polymorphism present for each of the eight microsatellites evaluated to vary substantially. Two microsatellites were found to be clearly associated with the presence of arcelin and the arcelin alleles in the germplasm survey. Of these, the microsatellite PV-ag004 was more polymorphic than PV-atct001. For both markers, unique alleles were found for the sources of arcelin compared to susceptible germplasm resulting in diagnostic tests for Arc1 and Arc5, the most commonly introgressed alleles of the arcelin gene [19].

Several microsatellites, notably PV-ag004 and PV-atct001, produced multiple banding, patterns especially in the wild beans which was interesting because these arcelin and phytohemagglutinin-based markers are part of sequences in the APA gene family which has several highly similar multiple copy sequences at least for the Arc5 [22]. It is interesting to note that microsatellites have been discovered in or near the arcelin and phytohemagglutanin members of the APA family, but none have been found in the alpha amylase gene. Additional multi-copy microsatellites have been found within a bacterial artificial chromosome clone from the Arc4 locus (pers. communic F. Sparvoli).

Linkage disequilibrium at the APA locus versus other regions of the genome

Given the large number of repeated sequences within the region of the APA locus we suspected LD might be low within this region however it was moderate, although it varied near and away from the region. For example among the markers PV-ag004 and PV-atct001 LD was high but then decayed towards markers BMd26 and BMd8. We found it interesting to contrast that decay of LD across these two groups of markers in wild versus cultivated Andean and Mesoamerican germplasm.

When wild and cultivated genotypes were evaluated separately, LD was relatively high in the wild genotypes both among the APA markers (r² = 0.4638) and across APA and non-APA markers from the same linkage group region (r² = 0.2971 to 0.7216). In contrast, for the resistant cultivated lines, LD was high between the two markers (r² = 0.9163 for PV-ag004 and PV-atct001) but low between these markers and the markers BMd26 and BMd8 (r² from 0.1172 and 0.1911). This may reflect reduced recombination within the segment containing the APA locus due to introgression from the wild accessions into the cultivated background. The pedigrees of the cultivated lines containing the arcelin introgression generally involved from one to two backcrosses with a standard Andean (eg. G76) or Mesoamerican (eg. Talamanca) background although some early lines in the Andean group were from simple crosses and some later lines were from crosses of a fixed RAZ line with another advanced line such as XAN252, CAP3 or AND685. The wild beans had high LD in contrast to the results of Rossi et al. [33] perhaps because our wild beans were all of Mesoamerican origin as this was the source of the arcelin locus.

Furthermore, LD for the APA locus was lower in the Andean genepool than in the Mesoamerican genepool. This difference may be due to the effect of introgressing the APA locus into an Andean genetic background compared to a Mesoamerican background given that the source of the arcelin gene is in wild Mesoamerican beans. The higher LD within the APA locus than outside it suggests that the APA locus from the wild sources of arcelin is often inherited as a block when it is backcrossed into a cultivated genetic background. This may explain why linkage drag has been observed in the breeding program for bruchid resistance, where it is possible to obtain advanced lines with resistance from the wild sources but these tend to have associated deleterious characteristics, such as lower yield potential and poor architecture which are associated with wild accessions of common bean.

The arcelin locus was unique and interesting for this study because of the microsatellites embedded near the APA family members and because of its status as both a seed storage protein locus and for its involvement in preventing insect herbivory. In contrast to the region of the APA locus, average LD values were high in several regions of the genome including linkage groups B6, B7 and B9 for both genepools and B1 for Mesoamerican beans and B11 for Andean beans. High LD in these regions could be explained by reduced diversity on certain linkage groups associated with the domestication genes fin and ppd on B1 [42] or heavily selected genes for seed color and seed size, p and Phs on B7; while the high average LD on B9 could be due to the effect of Fr, a fertility restorer gene on this linkage group.

Linkage disequilibrium between alleles at different loci varies with different plant genomes depending on mating system and location within the genome; with self pollinating species generally having higher levels of LD than cross-pollinating species and certain regions having higher LD due to reduced recombination or effects of selection [30]. In this sense, common bean as an inbreeding species should have a lower level of LD decay than outcrossing species although this would vary for different subpopulations of wild and cultivated genotypes [33].

Variability in recombination can be due to chromosomal context and structure, so that greater crossing-over occurs in euchromatin segments in gene dense regions, compared to heterochromatin or gene-poor regions [43, 44]. Furthermore, recombination is generally higher within tandemly repeated gene families especially those at disease resistance gene analog loci [45] and perhaps at seed storage protein loci [46]. In this context, the findings of this study are interesting because the moderate LD found at the APA locus may reflect the levels of recombination within the repeated gene family and the similarity of many of the gene family members which are tandemly arrayed and clustered on linkage group B4 [22]. It would be interesting to determine the extent of LD around other tandemly repeated genes. On the other hand, recombination can be suppressed at an introgression event when parts of a distantly related genome are backcrossed into a common background as happens when plant breeding directs the selection of a gene from a wild relative into a cultivated background.

Association of bruchid resistance and the APA locus

The data of the germplasm survey confirm that the APA family is responsible for bruchid resistance observed in this study as associations between the percentage adult emergence and the markers from the APA locus and even those flanking this locus were very significant. These results agree with the first evaluations of arcelin and the APA locus by Cardona et al. [29], Fory et al. [16], Kornegay et al. [19] and Osborn et al. [47]. More recent results of Zambre et al. [17] with transgenic tepary bean transformed with arcelin and of Goossens et al. [6] with artificial arcelin-containing seeds question the role of arcelin in resistance but these results may be due to the effect of heterologous expression or reduced concentration of the seed protein. Alternatively, other genes outside the APA locus may be needed in combination with arcelin for resistance to be expressed. Supporting this hypothesis Nishizawa et al. [20] found that bruchid resistance could be lost with backcrossing of arcelin alleles into a susceptible genotype.

Quantitative data on Z. subfasciatus egg laying and adult emergence across all the genotypes showed that these two traits were not correlated, therefore the mechanism of resistance was not based on any antibiosis at the seed coat level but rather on antibiosis to the development of the insect within the seed as was known previously. Despite this wild beans generally had fewer eggs laid per seed because the seed of these genotypes is so much smaller than cultivated beans. It would be interesting to explore this further especially in different sized common or lima beans as some differences in seed coat preferences have been noted [48] and both female fecundity and the numbers of insects emerging per seed in susceptible bean lines have upper limits [49].

Meanwhile, the percentage adult emergence was correlated with the number of adults in the cultivated resistant genotypes but not in the wild resistant genotypes perhaps because of an association with the number of eggs laid or because of variability for level of resistance found in the wild accessions. It was notable that the most useful APA variants in our study were Arc1 in the cultivated background and Arc1, Arc4, Arc5 and Arc7 in the wild background and this is consistent with results from previous studies [19]. Genetic background effect from wild germplasm background was also observed by Cardona and Kornegay [2] for the first five arcelin alleles and in the transgenic testing of arcelin variants described above. Arc1 has been used to a greater extent in plant breeders' selections as it provided higher resistance in a wide array of cultivar types of different seed sizes or colors compared to Arc5 or the other wild-derived alleles.

Arc2, Arc3 and Arc4, on the other hand, have not been widely used for introgression, so creation and analysis of cultivated genotypes with these alleles would be of interest with the provision that they are compared to Arc1 or Arc5 isolines. Finally, some resistant genotypes had very few adults emerge before 50 days after infestation but then had a few emerge in the period from 50 to 70 days after infestation. This may indicate that in resistant genotypes, especially those with weaker arcelin alleles, emergence may be delayed but a few adults do develop to maturity indicating the effect of antibiosis over the extended lifestages of the insect. It may be useful to seek recombinants between different Arc alleles to determine if various members of the APA family are epistatically influencing the expression of bruchid resistance.