The mapping population consisting of 200 lines was developed by crossing Mf10_2870, smut susceptible (J. Baldessari, unpublished data), and Ascasubi Hispano, smut resistant (Ibañez et al., 2018). An additional population used for validation purposes was also developed by crossing Granoleico, smut susceptible (Soave, 2002; Oddino et al., 2013) and Ascasubi Hispano. To rapidly identify QTL associated with smut resistance, a non-traditional approach was used where families within the population were phenotyped and genotyped at familial stages prior to the F₇ generation rather than developing a uniform recombinant inbred line (RIL) population. Briefly, three individuals from each F_2:3 family (for the mapping population) and from each F_3:4 family (for the validation population) were phenotyped under heavily smut infested field conditions. This was done during the 2019-2020 season for the mapping population and during the 2020-2021 season for the validation population, in General Deheza, Córdoba Province, Argentina under rainfed conditions.

Smut resistance tests were conducted as described in Chamberlin et al., 2022, in fields adjacent to a farm (-32.759, -63.770) in the town of General Deheza (Córdoba Province, Argentina). Soil at the site was a General Deheza coarse-silty sandy loam type (0.3% slope, coarse-silty, mixed, thermic, Entic Haplustoll). The tests were planted one km downwind from a peanut processing plant where peanut smut is prevalent. The production field site was on a rotation schedule of corn-corn-soybean-peanut, where the preceding crop of the test site was always soybean.

Two control entries, one resistant (Ascasubi Hispano) and one susceptible (Colorado Irradiado INTA) were planted in both seasons. They were replicated twelve times each. Each family was represented by three individuals, each one constituting a plot. Plots were arranged in a square grid and randomized within it. Controls were placed, alternately, along a diagonal of the grid. Individual plots were spaced one m apart among rows and between plants.

The tests were planted on 11/7/19 and 12/23/20, using an augmented grid design with three replications, using single plants as experimental units. Plants were respectively dug at 147 and 140 days after planting (DAP), and all pods from each plant were harvested by hand and placed in mesh bags to air dry for two months. Yearly soil spore count was done as described by Chamberlin et al. (2022) to determine disease pressure. The test was conducted following the extension guidelines of the Instituto Nacional de Tecnología Agropecuaria (INTA).

Each pod was opened by hand and rated for disease incidence, i.e., presence or absence of T. frezzii sori or spore masses on the kernels. Disease incidence (DI) was calculated as DI = Infected pods / Total pods.

DNA was taken from 572 individuals (approximately 3 individuals from each of the 200 lines) of the F_2:3 family (Mf10_2870 and Ascasubi Hispano. DNA was also taken from 96 individual plants of F_2:3 family of Ascasubi x Granoleico selected for phenotype. DNA extraction was performed from 20 mg of dried young leaves in silica gel. The samples were ground for two 10-sec cycles in bead mill at 15.000 rpm (Super FastPrep-2 Bead Beating System, MP BiomedicalsLLC, Irvine, CA, USA). A modified CTAB method was used with a sorbitol cleaning wash before the lysis step (Inglis et al. 2018). DNA quality was determined using 0.8% agarose gel and the DNA quantity was estimated with DS-11+ Nano UV-Vis Spectrophotometer (DeNovix Inc., DE, USA).

For whole-genome sequencing, seeds of Ascasubi INTA Cv, Granoleico Cv and Mf10_2870L were germinated in an incubation oven, under total darkness at constant temperature of 28 °C for 15 days. Fresh young etiolated leaves were collected for each genotype, immediately frozen with liquid nitrogen, and preserved at -80 C (Revco, Model ULT1386-5V41). The nucleii were extracted following the Nucleii Isolation – LN2 Plant Tissue Protocol (Ciculomics Inc./ PacBio) optimized from Workman et al. (2018). In the step 5 of protocol, filtration of the lysate with steriflip 20 µm pore was replaced with two lysate filtrations with a cell strainer of 100 µm pore and 40 µm pore. High-molecular weight (HMW) DNA was extracted from nucleii following the specifications of Nanobind Plant Nuclei Big DNA Kit (Circulomics Inc./ PacBio). DNA quality was determined using 0.8% agarose gel and the DNA quantity was estimated by Qubit V1 (Invitrogen ThermoFisher Scientific Inc.) using Qubit® dsDNA BR Assay Kits.

Seeds from the peanut mini core accessions were sowed in 4.7-m pots with a mix of 50% Promix (Premier Tech Horticulture, Quaker, PA) and 50% steam-sterilized sandy soil from the Coastal Plain Experiment Station in Tifton, GA. Young leaf tissue was collected from 1-month-old plants for DNA extraction using DNeasy Plant mini kit (www.qiagen.com). Quantification of DNA was performed with Quant-iT dsDNA assay kit (www.thermofisher.com). DNA were submitted for genotyping with the Arachis version SNP array consisting of 47K SNPs features (www.thermofisher.com). SNP data were curated from the Axiom analysis suite (www.thermofisher.com). SNP markers were classified into six categories, i.e., PolyHighResolution, NoMinorHom, MonoHighResolution, CallRateBelowThreshold, OfftargetVariant, and Other according to the SNP QC matrix of the software (Clevenger et al., 2017). All data analysis in this study was performed with markers in the PolyHighResolution category due to its high quality in signal separation.

High molecular weight DNA was received from A. hypogea Granoleico and A. hypogea Ascasubi. DNA was sheared using the Diagenode Megaruptor 3 targeting 20kb fragments. Sheared DNA was prepared for PacBio sequencing using the SMRTbell Express Template Prep Kit 2.0. The library was size selected with AMpure PB beads to remove fragments less than 3kb. Sequencing was performed on a Sequel IIe System (Pacific Biosciences, Menlo Park, CA) using Binding Kit 2.2, Sequel II Sequencing Kit 2.0, and SMRTCell 8M. To target HiFi reads, the library was sequenced using a 30-hour movie time using Instrument Control Software Version 10. Raw subreads were converted to HiFi data by processing with CCS to call a single high quality consensus sequence for each molecule, using a 99.5% consensus accuracy cutoff. A. hypogea Granoleico was sequenced on 2 SMRTcells yielding 73.11Gb of HiFi data, and A. hypogea Ascasubi was sequenced on 3 SMRTcells yielding 61.37Gb of HiFi data.

Genotyping: Khufu whole-genome sequencing and analysis

DNA from populations were prepped into sequencing libraries using iGenomX Riptide library prep. For the F2:3 population, Six Riptide libraries, representing 576 samples were sequenced on 2 lanes of NovaSeq S4 chemistry (800 Gb raw base pairs per lane) to yield an average of 0.99X genome coverage per individual. The raw reads were demultiplexed and processed using Khufu (hudsonalpha.org/khufudata). To identify QTL for resistance, bulks were identified using one year of field data. Analysis of the bulks were done in silico, by bulking individual sequenced samples and calculating allele frequency differences between the putative resistant and susceptible bulks. The F3:4 samples were processed and analyzed the same way, except were sequenced to an average of 1.14X genome coverage. Haplotype analysis was done using HawkHAP, which is a component of Khufu.

Long read sequencing and assembly led to the generation of highly contiguous genome assemblies for smut resistant parent Ascasubi, and smut susceptible parent Granoleico (Table 1).

A total of 576 individuals to construct bulks for mapping. They represented 200 F2:3 sub-families where each family was represented by 3 individuals. The smut pressure was low due to drought stress during pod development, but there were enough ‘susceptible’ individuals to map construct bulks (Figure 1). The frequency of infection was heavily skewed to the low end of the distribution, indicating that there was a considerable amount of false negative disease scores. To construct bulks, we employed a conservative approach whereby we selected susceptible individuals when all three sub-family members had infection and the average infection was greater than 5%. It was more difficult to select resistant bulks, because more than 400 individuals had 0% infection. For the ‘resistant’ bulk we selected every family member if they each had 0% infection and at least one individual was located in the field close to a susceptible check that exhibited infection. These criteria were chosen to reduce false negatives in the bulk as much as possible.

We selected 45 individuals for the ‘resistant’ bulk and 49 individuals for the ‘susceptible’ bulk. The QTL-seq analysis was conducted by bulking bam alignment files in silico and assaying allele frequency differences genome-wide between the bulks (Figure 2A). Analysis of the bulks indicated one major QTL located on the proximal end of chromosome 12. There were no other peaks identified, indicating that a single, major locus accounted for smut resistance from Ascasubi. In the following year (2019-2020), we phenotyped a related population of F3:4 progeny with Ascasubi as a parent and Granoleico as the susceptible parent. We selected a set of 48 resistant individuals and the 48 most susceptible individuals. We sequenced those 96 individuals and again analyzed them with QTL-seq analysis. The major locus on chromosome 12 was again indicated as a single, major resistance locus controlling resistance (Figure 2B).

We have identified and validated a major locus controlling smut resistance in peanut. The qualitative inheritance of this trait, its high level of heritability, low environmental variance, and strong phenotyping methodologies, we were able to map and validate this resistance within early generations (F2:3 and F3:4), by phenotyping single plants rather than replicated plots. Phenotyping and genotyping early generation families within a developing population saves time and allows the potential for selection for backcross breeding. This method greatly decreased the time needed for discovery. The resistance is indicated on an approximal 2 Mb region that contains the largest cluster of R genes in the peanut genome (Bertioli et al., 2016). We have generated highly contiguous genome assemblies of the resistant parent, Ascasubi, and of the most widely grown cultivar in Argentina, Granoleico. Granoleico is very susceptible to smut, and the differences between those two genomes within the 2 Mb region will contain the functional variation that confers strong resistance to smut. This resistance and the markers associated with it will be transformational in breeding smut resistant cultivars for Argentina and the world.