**1. Introduction**

Rice is one of the most important staple crops for more than half of the population in the world [1]. Rice blast, caused by the fungus *Magnaporthe oryzae*, is one of the most devastating diseases of rice, causing yield losses of 10%–30% annually [2]. The development and use of resistant varieties appears

to be the most economical and environmentally sustainable way to control rice blast [3]. Identification of genes or genetic loci conferring resistance to blast disease could accelerate breeding programs for resistant rice varieties.

Genetically, disease resistance in plants can be categorized into two types, qualitative and quantitative [4,5]. Qualitative resistance is mainly mediated by a single resistance gene (*R* gene), which confers complete but race-specific resistance through the recognition of pathogen effectors [6,7]. *R* gene-mediated resistance has been widely deployed in crop breeding programs. However, *R* gene-mediated resistance is often not durable, as most pathogens are able to rapidly evolve new virulent races lacking the corresponding avirulence effectors to evade recognition by the cognate R protein. Quantitative resistance is mediated by multiple genes or quantitative trait loci (QTLs), providing partial or basal resistance associated with delayed and reduced development of disease lesions [8,9]. Although quantitative resistance has only partial effects, it has been considered race-non-specific and broad-spectrum, and is therefore of particular interest for breeding crops with durable resistance [4,10].

To date, more than 100 rice blast *R* genes have been identified and at least 28 *R* genes have been cloned [11,12]. All the cloned major *R* genes encode nucleotide-binding site leucine-rich repeat (NBS-LRR) proteins, with the exception of *Pid2*, encoding a B-lectin kinase [13], and *Ptr*, encoding an Armadillo repeat protein [14]. Several hundred QTLs associated with blast resistance have been identified [15]. However, only a limited number of QTLs for blast resistance have been cloned [16–20]. The cloned QTLs encode proteins that are diverse in their structure and function: *pi21* encodes a proline-rich protein with loss-of-function deletions [16]; *Pb1* encodes an atypical coiled-coil (CC)-NBS-LRR protein [17]; *Pi35* and *Pi63* encode NBS-LRR proteins [18,19]; whereas *bsr1-d1* encodes a C2H2-type transcription factor with a single nucleotide change in the promoter [20]. These findings indicate that quantitative resistances are controlled by diverse molecular mechanisms.

With the rapid development of next-generation sequencing, approaches based on bulked segregant analysis coupled with whole-genome sequencing (BSA-Seq) have been developed for the mapping of agronomically important loci in rice [21–23], including major genes or QTLs responsible for blast resistance [22,24,25]. In the present study, we apply BSA-seq to rapidly map four QTLs, *qBBR-4*, *qBBR-7*, *qBBR-8*, and *qBBR-11*, responsible for basal resistance to blast disease, and identify a novel haplotype of the durable blast resistance gene *pi21* as a candidate gene of *qBBR-4* on chromosome 4 in a *japonica* variety 02428 (*pi21-2428*). While the resistant *pi21* gene was found only in *japonica* before [16], we identify three Chinese *indica* varieties carrying the resistant *pi21-2428* allele in 325 accessions. Therefore, the novel *pi21-2428* allele and the *pi21-2428*-containing rice varieties identified in the present study provide valuable resources for breeding rice varieties, especially *indica* rice, which are durably resistant to blast disease. Our results also lay the foundation for further identification and functional characterization of the other three QTLs for a better understanding of rice basal resistance to blast disease.

### **2. Results**

#### *2.1. Evaluation of 02428 and LXG in Basal Resistance to Rice Blast Disease*

In our earlier evaluations, the rice variety 02428 was observed to possess high basal resistance to the rice blast fungus *M. oryzae* under natural nursery conditions (data not shown). We further performed artificial inoculations on 02428 seedlings using three virulent isolates of *M. oryzae* in this study. The results showed that 02428 was moderately susceptible to isolates 501-3 and Guy11, and was moderately resistant to isolate RB22 (Figure 1A). Most lesions on leaves of 02428 were limited in size. In contrast, LiXinGeng (LXG) was highly susceptible to all three isolates. These results suggest that 02428 possesses high basal resistance, preventing blast disease development.

An F2 population of 02428 × LXG with 626 individuals was inoculated with RB22. The results show that the frequency distribution of disease severity in the F2 population of 02428 × LXG exhibited continuous variation (Figure 1B), indicating that the resistance to RB22 is likely controlled by multiple genes.

**Figure 1.** Resistance reaction of rice varieties 02428, LiXinGeng (LXG) and their F2 population to rice blast disease. (**A**) Phenotypes of 02428 and LXG inoculated with *M. oryzae* isolates 501-3, Guy11, and RB22. (**B**) The frequency distribution of disease severity in the F2 population of 02428 × LXG inoculated with *M. oryzae* isolate RB22. Disease severity was assessed following a 0–5 scale (0–1: resistant, 2: moderately resistant, 3: moderately susceptible, 4–5: severely susceptible).

#### *2.2. SNP and Short InDel Polymorphism Profiling*

Whole-genome sequencing of extremely resistant (ER) and extremely susceptible (ES) pools derived from the F2 population of 02428 × LXG and the two parental lines 02428 and LXG generated about 90.7 to 158.9 million reads for each sub-pool or parental line (Supplementary Table S1). After filtering, a total of 469,512 bi-allelic single-nucleotide polymorphisms (SNPs), and a total of 65,766 bi-allelic short insertions and deletions (InDels) were identified (Table 1). The average densities of SNP and short InDel markers were about 1.26 SNP/kb (average every 795 bp exists a SNP) and 0.18 InDel/kb (average every 5,675 kb exists a short InDel) (Table 1), respectively. The polymorphic markers were sufficiently distributed across the whole genome, except for one region of about 6.5 Mb on chromosome 3 containing relatively fewer markers (Supplementary Figure S1).


**Table 1.** Chromosome-wise distribution of the identified single-nucleotide polymorphisms (SNPs) and short InDels.
