Genetic Dissection of the Breakdown of Durable Resistance in Indica Rice Variety PTB33 to Brown Planthoppers Nilaparvata Lugens (Stål)


 BackgroundTo avoid and delay the resistance breakdown of varieties against pathogens and insect pests, broad-spectrum and durable resistance by multiple genes pyramiding are expected to be one of the practical approaches. The indica rice variety PTB33 (Oryza sativa L.) shows high and durable resistance to the brown planthopper (BPH, Nilaparvata lugens Stål). However, this variety gradually lost its resistance against the recent virulence development of BPH. However, breakdown processes are not fully elucidated by individual genetic loci. ResultsEffective QTLs were explored across the whole genomic region against four BPH populations collected in Japan in 1988, 1989, 1999, and 2013 using high-density single-nucleotide polymorphism (SNP) markers obtained by genotyping-by-sequencing. Among seven genomic regions of PTB33 likely conferring BPH resistance, four QTLs, qFSA4a, qFSA6, qFSA11, and qFSA12 on chromosomes 4, 6, 11, and 12, respectively, were validated as BPH resistance QTLs. The PTB33 alleles at the four QTLs positively contributed to BPH resistance. Infestation of monogenic segregating lines showed that the PTB33 alleles at qFSA11, qFSA12, and qFSA4a lost resistance effects at least in 1989, 1999, and 2013, respectively. ConclusionThis study showed breakdown of durable resistance in PTB33 was explained by step-by-step losses of genetic effects at each resistance locus and probably multiple acquisitions of virulence genes in BPH in a gene-by-gene specific manner.

species are unable to overwinter in Japan and migrate to Japan each year from the Chinese mainland during the early part of the rice-growing season (Kisimoto 1976). The biotypes and virulence changes of the BPH population in Japan are likely to be affected by continental East Asian BPH populations (Reviewed by Matsumura 2001).
To date, more than 40 BPH resistance genes have been reported in rice varieties and wild species (Fujita et  However, varieties carrying a single resistance gene exhibit rapid decay of resistance within a few years after release owing to spatial and temporal divergence in the virulence of BPH populations. IR26, the rst variety with the resistance gene BPH1, was developed in 1973, and its cultivation has been widely promoted in Southeast Asian countries. However, in 1975, a loss of resistance was reported (Cohen et al. 1997). Although the variety IR36 harboring the BPH2 resistance gene was released in 1976 following the release of IR26, its resistance was not effective in the early 1980s, probably due to the changing virulence of BPH populations (Khush and Virk 2005). These results indicate that introducing only a single resistance gene permits rapid adaptation of BPH to the resistance gene, making it di cult to achieve a sustainable effect. However, it has also been reported that varieties integrating multiple resistance genes show a more enhanced and durable resistance . Therefore, it is necessary to evaluate how long individual genes in durable resistance variety are effective to BPH.
The indica rice variety PTB33 (O. sativa L. ssp. indica) shows wide-spectrum resistance to BPH populations with different virulence patterns in South and Southeast Asia, including India, Taiwan, Vietnam, Indonesia, and the Philippines (Horgan et al. 2019). The BPH resistance of PTB33 can be explained by the presence of at least three loci, BPH32, BPH2, and BPH17. BPH32 on chromosome 6 has been isolated using a map-based cloning approach. BPH2 has been identi ed using linkage mapping (Angeles et al. 1985;Ren et al. 2016). In addition, PTB33 possesses an identical amino acid sequence to the BPH resistance gene BPH17 on the distal end of the short arm on chromosome 4, which was isolated from the indica rice variety Rathu Heenati . Three near-isogenic lines (NILs) carrying chromosome segments derived from PTB33 at BPH32, BPH2, and BPH17 exhibited higher resistance effects in the genetic background of the japonica rice variety Taichung 65 (T65) ). However, these three genes could only partially explain the high-level resistance of PTB33. Therefore, genome-wide explorations of genes or quantitative trait locus (QTL) with other additive and epistatic genetic effects would be necessary better to understand the durable resistance and its breakdown of PTB33.
Here we conducted genetic dissection of breakdown processes of BPH resistance in PTB33 by individual genetic loci by genome-wide QTL analysis using high-density SNP markers obtained by genotyping-bysequencing (GBS) with the infestation of the four BPH populations collected in Japan in 1988,1989,1999, and 2013.  (Table S1). The BPH resistance of PTB33 was evaluated by the rate of females with a swollen abdomen (FSA) at ve days after infestation (DAI) by the 1966-Hadano, 1989-Chikugo, 1999-Koshi, and 2013-Koshi BPH populations were 0.0%, 4.0%, 16.0%, and 55.0%, respectively, suggesting that the virulence of BPH populations increases with the collection year of BPH ( Fig. 1).

QTL estimation for BPH resistance
The effective resistance QTLs to the four BPH populations were investigated in hybrid progenies derived from a cross between the susceptible varieties T65 and PTB33 (Fig. S1). The In the PTB33 backcrossed population, the FSA of the 1966-Hadano, 1989-Chikugo, 1999-Koshi, and 2013-Koshi BPH populations were 12.1%, 11.3%, 14.6%, and 34.5% at 3 DAI ( Fig. 2a-d)  Multiple QTL mapping, including qFSA4a as a covariate, detected the 2 nd QTL, qFSA6, at 2.2 Mbp on chromosome 6. Several cycles of multiple QTL mapping, including all detected QTLs as covariates, were repeated until the updated genetic model did not improve the LOD value by more than 3.0 in the likelihood ratio test (see Methods). Collectively, multiple QTL mapping for FSA detected ve QTLs (Table 1): qFSA4a, qFSA6, qFSA11 at 24.3 Mbp on chromosome 11, qFSA7a at 1.6 Mbp on chromosome 7, and qFSA3 at 10.3 Mbp on chromosome 3. The genetic effects of the ve QTLs were estimated simultaneously: qFSA4a had the largest genetic effects, with 43.0% PVE and -28.8% additive effects; qFSA6 explained 6.5% of the PVE and an additive effect of -13.9%; qFSA11 had a PVE of 5.0% and an additive effect of 16.9%; qFSA7a had a PVE of 4.3% and an additive effect of -15.8%; and qFSA3 had the lowest PVE of 4.0% and an additive effect of -9.6%.
Next, effective QTL to the 1989-Chikugo BPH at 5 DAI was explored. All of the QTLs detected by the 1966-Hadano were not detected. Instead, qFSA4b at 13.0 Mbp on the short arm of chromosome 4 and qFSA12 at 23.5 Mbp on the long arm of chromosome 12 were involved in BPH resistance (Table 1). qFSA4b explained a PVE of 20.6% and an additive effect of -20.8%. qFSA12 explained a PVE of 11.9% and an additive effect of -15.8%. In the BC 1 F 3 population infested by the 1999-Koshi BPH population, one QTL with a PVE of 22.6% and an additive effect of -15.3% was detected at 7.4 Mbp on chromosome 4, which seemed equivalent to qFSA4a (Table 1). No QTLs were detected in the BC 1 F 3 population infested by the 2013-Koshi BPH population.
In the PTB33 backcrossed population, no QTLs were detected for FSA at 5 DAI in any BPH population, probably due to small phenotypic variance because almost segregating plants showed strong resistance. However, at 3 DAI, several QTLs were successfully detected ( Table 2). In the BC 1 F 3 population infested by the 2013-Koshi BPH population, qFSA6 at 2.4 Mbp on chromosome 6 and qFSA8 at 18.0 Mbp on chromosome 8 were involved in BPH resistance with additive effects of -25.7%. and -14.6%, respectively).

QTL identi cation in T65 genetic background
The suggested QTLs for FSA were identi ed using anking SSR markers by the infestation of the 1966-Hadano BPH population at 5 DAI. In the B 1 F 3 population derived from backcrossing with T65, plants heterozygous at the QTL and homozygous for T65 or PTB33 at the background QTL were self-pollinated to obtain B 1 F 4 populations as the monogenic segregating populations (Fig. 3, Table S3). LOD peaks at qFSA4a, qFSA6, qFSA11, and qFSA12 exceeded over more than a 5% empirical threshold in the corresponding MSP: qFSA4a was repeatedly detected in B 1 F 4 population #102, and the PTB33 allele was estimated to reduce 33.2% of FSA at 5 DAI; qFSA6 was identi ed in population B 1 F 4 #114, with an additive effect of -26.5% for the PTB33 allele. Similarly, qFSA11 and qFSA12 were identi ed with additive effects of -13.2% and -11.5% for the PTB33 allele in B 1 F 4 populations #109 and #110, respectively. The remaining three QTLs (qFSA3, qFSA4b, and qFSA7a) were not identi ed in these populations.

Assessment of genetic effects at QTLs responding to insect population
The reduction of genetic effects at the qFSA4a, qFSA6, qFSA11, and qFSA12 to developing virulence of the four BPH populations was statistically measured by LOD i , LOD scores attributed from interaction component between QTLs and BPH populations ( Table 3). The phenotypic values and genotypes of the B 1 F 2 populations infested by 1966-Hadano BPH and the B 1 F 3 populations infested by 1999-Koshi BPH were mixed and simultaneously solved to estimate LOD i . The LOD i score at qFSA4a was higher than the 5% signi cant threshold, implying that genetic effects at qFSA4a were different by 1966-Hadano and 1999-Koshi BPH populations. This difference is likely due to the shrinkage of the additive effect of FSA from -28.8% in the 1966-Hadano BPH to -15.3% in the 1999-Koshi BPH (Tables 1 and 3).
The LOD i scores at qFSA11 did not show apparent at 5% signi cance level but showed signi cance at 10% level in population size in this study. The additive effects of the PTB33 allele at qFSA11 were -16.9% or zero (null hypothesis) by the 1966-Hadano or 1999-Koshi BPH, respectively. The LOD i scores at qFSA12 implied that genetic effects were different by the BPH populations at a 10% signi cance level. The additive effect at the qFSA12 was signi cantly estimated with -11.5% for the PTB33 allele in the B 1 F 4 monogenic segregating population infested by the 1966-Hadano BPH infestation (Table S3). However, qFSA12 was not detected in the B 1 F 2 generation, probably because residuals were not su cient enough to detect qFSA12 under simultaneous estimation with qFSA4a, qFSA6, qFSA11, qFSA7a, and qFSA3 in this population size. On the other hand, the interaction of qFSA6 by the BPH populations was not apparent, while qFSA6 conferred resistance by the infestation of the 1966-Hadano BPH but not to the 1999-Koshi BPH population ( Table 3).
The genetic effects of the PTB33 alleles at the qFSA4a, qFSA6, qFSA11, and qFSA12 to the four BPH populations were clari ed the B 1 F 5 plants homozygous for T65 (T) and PTB33 (P) were obtained from a single heterozygous MSP of each QTL (Fig. 4) and infested by the four BPH populations. Female insects with medium swollen abdomens (MS) and non-swollen abdomens (NS) were evaluated as those with reduced growth and insects with swollen abdomen as those with normal (N) growth. Infestation with 1966-Hadano BPH revealed an apparent resistance effect of qFSA4a, qFSA11, and qFSA12 on homozygous plants for PTB33, as compared with homozygous plants for T65 in Fisher's exact test. However, the resistance effect of the PTB33 allele at qFSA11 was lost in the 1989-Chikugo insect population, while the other QTLs qFSA4a and qFSA12 were still effective. The PTB33 allele at qFSA12 lost resistance effect, whereasone at qFSA4a remained to the 1999-Koshi BPH population. However, all four QTLs lost resistance to the 2013-Koshi insect population. Therefore it was concluded that the PTB33 alleles at qFSA11, qFSA12, and qFSA4a lost resistance effects in 1989, 1999, and 2013, respectively.

Discussion
This study detected four QTLs conferring BPH resistance in backcrossed progenies between a resistant variety PTB33 and a susceptible variety T65 infested by four BPH populations with developing virulence by year in Japan. It has been reported that varieties harboring a single resistance QTL/gene with major genetic effects rapidly promote genetic changes or adaptations of BPH populations and result in resistance breakdown of resistant variety. In contrast, a pyramiding of multiple resistance genes enables resistance to BPH to persist for a longer duration . Multiple BPH resistance genes might cause resistance of PTB33 to BPH populations from most South and Southeast Asian countries (Horgan et al. 2015). However, the BPH resistance of PTB33 gradually reduced across years (Fig. 1). Nevertheless, PTB33 was still resistant to the recently collected BPH populations (2013-Koshi BPH population). These results suggest that the persistent resistance of PTB33 to BPH is mediated by the presence of multiple BPH resistance genes.
We focused on the QTL conferring FSA at 5 DAI in the T65 backcrossed population (Table 1). qFSA4a, qFSA6, qFSA11, qFSA7a, and qFSA3 were detected in the infestation of the 1966-Hadano BPH population. qFSA4b and qFSA12, which were not detected in the 1966-Hadano BPH population, were detected in the infestation of the 1989-Chikugo BPH population. In addition, qFSA4a was detected in the infestation of the 1999-Koshi BPH population. No QTLs were detected in the infestation of the 2013-Koshi BPH population. These suggest that PTB33 may have changed the QTL conferring BPH resistance in the four BPH populations. In the PTB33 backcrossed population, qFSA6 and qFSA8 were detected in the infestation of the 2013-Koshi BPH population at 3DAI (Table 2). qFSA8 was only detected in the PTB33 backcrossed population, likely due to interaction with other QTLs in the genetic background of PTB33. Identi cation of qFSA7b and qFSA8 is necessary for further analysis in the appropriate genetic backgrounds. To date, most of BPH resistance loci were contained in a cluster of four chromosomal regions designated clusters A, B, C, and D on the long arm of chromosome 12, the short arm of chromosome 4, the short arm of chromosome 6, and the long arm of chromosome 4, respectively (Fujita et al. 2013). qFSA4a, qFSA6, and qFSA12 were detected in clusters B, C, and A, respectively. qFSA4a, qFSA6, and qFSA12 were inferred to be identical to BPH17, BPH32, and BPH2, respectively, on previous genetic studies using PTB33. qFSA11 is a novel locus of BPH resistance.
Individual genetic loci in this study dissected the resistance breakdown of PTB33. Since BPH are unable to overwinter, the biotypes and virulence changes of BPH are likely to be affected by continental East Asian BPH populations in Japan (Matsumura 2001). The PTB33 allele at qFSA12 lost resistance between 1989 and 1999 in Japan (Fig. 4). qFSA12 was located on Fujita's cluster A containing BPH1 and BPH2. The BPH immigrating into Japan became virulent to a resistance gene BPH1 around 1988-1990 (Sogawa, 1992), and the BPH population rapidly became virulent to BPH2 beginning in 1997 (Tanaka and Matsumura 2000). Therefore, qFSA12 is likely to be identical to BPH1 or BPH2 corresponding to the previous reports. A recessive virulence gene to BPH1, vBPH1, was identi ed on the linkage map of BPH, indicating BPH resistance resulted from BPH1 occurred in a gene-for-gene manner with the resistance gene in a host rice plant and the avirulence gene in a pest insect BPH (Kobayashi et al. 2014). The PTB33 allele at qFSA4a was found to lost resistance between 1999 and 2013 (Fig. 4). Horgan et al. (2015) reported that a single BPH resistance gene was insu cient to resist the current BPH populations increasing their virulence. The evolution of the virulence gene would explain the recently increasing virulence to at least qFSA4a or BPH17. The PTB33 allele at qFSA11 would lose resistance as early as 1989 in Japan, although a BPH resistance gene has not been detected around this region. In the initial breakdown process to the BPH resistance gene between 1966 and 1989, not only BPH1 and BPH2 but also other unknown genetic loci, including qFSA11, might have already lost their resistance effect.

Conclusion
This study suggested that reduced genetic effects at each resistance locus explain the breakdown of durable resistance in PTB33 due to acquiring multiple virulence genes in BPH. The virulence mechanisms of BPH need to be clari ed in a gene-by-gene manner using BPH resistance gene NILs in further studies.

Plant materials
A susceptible japonica variety, T65, and a resistant indica variety, PTB33, were used as female and male parents, respectively, to produce F 1 hybrids. The F 1 plants were backcrossed with T65 or PTB33 to produce the B 1 F 1 population (Fig. S1) Table  S1). The collected populations were continuously reared using the japonica variety Reiho at the Kyushu-Okinawa Agricultural Research Center (KARC), Koshi City, Kumamoto Prefecture, Japan, at 25°C for 16 h under light and 8 hours dark cycle conditions. BPH resistance was evaluated using the 1966-Hadano BPH population in the B 1 F 2 , B 1 F 4 , and B 1 F 5 populations and the remaining four BPH populations in the B 1 F 3 and B 1 F 5 populations (Fig. S1).

Evaluation of BPH resistance
The antibiosis effects of rice on BPH were determined by assessing female adult mortality (FAM) and the rate of females with a swollen abdomen (FSA) of BPH. First, a single seed was sown in a 215 ml plastic cup and grown for approximately 80 days. Five adult females within 48 h after emergence were used to infest a single plant in a transparent tube. FAM and FSA on three and ve days after infestation (3 DAI and 5 DAI) were examined as the percentage of dead insects and the percentage of insects with swollen abdomens, respectively, of the total insects on a single plant. The degree of FSA was evaluated at three levels: massively swollen abdomen (S) if extremely large, non-swollen abdomen (NS) if a swollen abdomen was not observed, and a medium swollen abdomen (MS) if a moderately swollen abdomen was noted between S and NS.

Genotyping-by-sequencing
For extraction of total genomic DNA, 20-40 mg of young leaf samples frozen in liquid nitrogen were ground in a 2.0 ml plastic tube using a Multibeads shocker (Yasui Kikai, Osaka, Japan). Total genomic DNA was extracted using the protocol described by Mayjonade et al. (2016). 10 mg of genomic DNA digested with the restriction enzymes KpnI and MspI was used for library preparation of GBS (Poland et al. 2012). Library sequencing was performed using the MiSeq platform (Illumina, San Diego, CA, USA).
Short read sequences (in .fastq format) of the GBS library were processed using the Tassel5 software (Bradbury et al. 2007) to conduct genotyping calling. Imputation of missing genotypes or incorrect genotype correction was conducted using a hidden Markov model with the low-coverage biallelic impute (LB-Impute) software (Fragoso et al. 2016).

QTL analysis
Genome-wide LOD values were scanned using marker regression with R/qtl (Broman et al. 2003). The LOD threshold at the 5% signi cance level was determined based on 1000 permutation tests. The DNA marker with the highest signi cant LOD score was included in the genetic model, and another QTL was explored under the assumption that multiple additive QTLs simultaneously control genetic values without interactions in multiple QTL mapping (MQM). When a new marker increased the LOD value to 3.0, the added genetic locus was accepted as a QTL.

Genotyping by SSR markers
Total genomic DNA was extracted according to the method described by Dellaporta et al. (1983), with minor modi cations. Primer sequences of the polymerase chain reaction (PCR)-based markers used in this study are listed in Table S2. Each 15-mL reaction mixture consisted of 50 mM KCl, 10 mM Tris (pH 9.0), 1.5 mM MgCl 2 , 200 mM dNTPs, 0.2 mM primers, 0.75 units of Taq polymerase (Takara, Otsu, Japan), and 10 ng genomic DNA template. PCR was performed using the GeneAmp PCR System 9700 (Applied Biosystems, Foster City, CA, USA). The cycling pro le consisted of an initial denaturation step at 95 °C for 5 min; 35 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 40 s; and a nal elongation step at 72 °C for 7 min. Ampli ed products were electrophoresed on a 4% agarose gel in 0.5× TBE buffer.

Interaction between QTL and BPH population
Interactions between QTL and BPH populations were calculated as described by Broman et al. (2009).
Brie y, the 1966-Hadano and 1999-Koshi BPH populations were coded as 0 and 1 as covariates for input into the linear models. The effect of BPH populations was partitioned by the additive effect and interaction between the QTL and BPH populations. We assumed hypotheses of the null model (H 0 ), the additive model (H a ) assuming a single QTL and BPH populations, and the full model (H f ) including the additive effect of a single QTL and BPH, and interaction between the QTL and BPH. The log 10 likelihood ratio (LOD) comparing H f and H 0 , LOD f , represents evidence for the QTL, BPH, and interaction. The other LOD score, LOD a , comparing H a and H 0 provides evidence for the additive effects of QTL and BPH. LOD i is de ned as the difference between LOD f and LOD a , LOD i = LOD f -LOD a , implying evidence of interaction between QTL and the BPH population. The LOD threshold at the 5% signi cance level was determined based on 1000 times permutation test.

Availability of Data and Materials
The data sets supporting this article are included in the article and in the additional les. Short-read sequences of genotyping-by-sequencing are deposited to the DNA databank of Japan (DDBJ) Read Archive submission number DRA012495.
Ethics Approval and Consent to Participate

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementalTable.xlsx