First identified in Japan in 1884, BLB has been long recognized as a major threat in many Asian countries, and therefore for decades at the heart of many research and resistance breeding programs (Liu et al. 2014). To date, a handful of avirulence genes have been cloned from Asian Xoo strains, all of which encode TAL effectors including avrXa7, avrXa10, avrXa23, and avrXa27 (Hopkins et al. 1992; Gu et al. 2005; Wang et al. 2014). In contrast, no avirulence gene has been identified in African Xoo strains, the first of which were isolated in the 1980s. However, as many as nine races of African strains, not found among Asian strains, have been reported so far (Gonzalez et al. 2007; Tekete et al. 2020), indicating that some effector genes of African Xoo strains act as avirulence factors in gene-for-gene interactions with some rice R genes. Comparison between African and Asian Xoo TALomes shows that no tal gene is conserved between these two Xoo lineages (Lang et al. 2019). Strains of both lineages induce the S genes OsSWEET14 and OsTFX1 through unrelated TALEs, highlighting cases of inter-lineage evolutionary convergence of virulence functions (Streubel et al. 2013; Tran et al. 2018). In contrast, none of the rice E genes induced by Asian Xoo is predicted to be targeted by any African Xoo TALE. Here, toward identifying sources of resistance effective against African Xoo strains, we assessed the putative avr activity of 12 previously cloned individual tal genes representing the TALome of two African Xoo strains (Tran et al. 2018), using a large set of rice accessions.
To accomplish this goal, we first took a gain-of-function approach using the virulent Asian Xoo strain PXO99A as a recipient to express cloned African tal genes. In a similar approach, we previously identified major virulence TALEs of African Xoo strains using the US strain Xo X11-5A as recipient (Tran et al. 2018). X11-5A lacks tal genes and is weakly virulent, allowing for gain-of-virulence screening (Ryba-White et al. 1995; Triplett et al. 2011). Nonetheless in a search for gain-of-avirulence, the use of a more virulent strain is required. By selecting the Asian strain PXO99A as recipient, we also minimized the risk of TALE functional redundancy. This strategy led to the identification of talI and talF respectively causing resistance in rice varieties CT13432 and FKR47N, as well as talD and talH, each triggering resistance on IR64.
In a complementary approach, a loss-of-function analysis was carried out by mutagenizing individual tal genes in the African Xoo strain BAI3. This analysis validated talD and talI as avirulence genes; a mutant for talH was not obtained. Because IR64 has been sequenced and recombinant populations are available (Fragoso et al. 2017), prospects for future identification of the gene(s) underlying the TalD-elicited resistance are good. Mutagenesis of talI or talF failed to cause a measurable loss of avirulence in the respective rice varieties CT13432 and FKR47N. However, talI avirulence activity was evident upon trans-expression of the truncTALE tal2h in BAI3ΔtalI, revealing that a resistance like that mediated by Xa1 in response to TALEs generally and suppressed by truncTALEs, is present in CT13432 and masking the TalI-specific resistance. By further taking advantage of this combined gain- and loss-of-function approach, we also determined that the TalI-specific resistance depends on the activation domain of the effector and thus likely involves an E gene.
While four TALE groups including TalG, TalE, TalD and TalC are strictly conserved between Xoo strains MAI1 and BAI3, two to six RVD variations were reported in the five other groups, modifying to some extent their DNA binding sequence specificities (Tran et al. 2018). Notably, three of the four candidates or validated avirulence TALEs identified in this study, TalI, TalF, and TalH all present RVD polymorphisms across Malian and Burkinabe strains, which could be the result of some ongoing selection pressure imposed by corresponding E or other resistance genes (Doucouré et al. 2018; Schandry et al. 2018). Surprisingly, the TalD group is much more conserved, highlighted by an absolute RVD conservation between BAI3, MAI1 and the Cameroonian strain CFBP1947 (Tran et al. 2018), and only one polymorphic RVD among nine Malian strains investigated (Doucouré et al. 2018). According to the EBE prediction tools Target Finder and Talvez (Doyle et al. 2012; Pérez-Quintero et al. 2013), 90% of the first 20 predicted targets of the two versions of TalD are conserved. Such a conservation among strains hints to an important role in virulence for TalD, but there are no data supporting this hypothesis so far. Concerning the TalI group, BAI3 and MAI1 alleles differ at two RVDs and share about 32% of their predicted targets, querying the Nipponbare genome. Since both variants equally elicit resistance in the rice variety CT13432, their polymorphism may confer an advantage in future identification of the putative corresponding E gene; the gene is likely to reside among the relatively smaller number of predicted targets in that genome that are shared by the two alleles.
Up to now 12 R genes against BLB have been cloned, of which 9 are triggered by TALEs (Jiang et al. 2020; Chen et al. 2021; Luo et al. 2021). This proportion not only highlights the crucial role of this effector family in the rice-Xoo pathosystem but also the relevance of using TALEs as probes to identify resistance sources and clone the underlying genes. This quest is facilitated by the ability to identify candidate targets of TALEs by combining EBE prediction and expression analysis, owing to the modular nature of TALE DNA interactions and their strong transcriptional upregulation of targets (Boch et al. 2014). In a proof-of-concept study pioneering this strategy for E gene identification, Strauss et al. (2012) identified the pepper Bs4C gene, which is specifically induced by the TALE AvrBs4 from X. euvesicatoria. More recently, a mix of map-based cloning and EBE predictions led to the isolation of the E gene Xa7 in rice, which is triggered by the Xoo TALE AvrXa7 (Chen et al. 2021; Luo et al. 2021). In our study, out of the 86 accessions phenotyped, 12 were found to be resistant to both African Xoo strains MAI1 and BAI3, including 7 accessions of O. sativa, 2 of O. glaberrima and 3 NERICAs. Moreover, 12 O. glaberrima, 3 O. sativa, 3 O. barthii and 2 NERICA accessions were discovered to be resistant specifically to the Malian strain MAI1.
Among the 12 accessions identified as resistant to Xoo African strains MAI1 and BAI3, 9 accessions, including CT13432 and FKR47N, showed a phenotype typical of Xa1 resistance (Additional file 6: Table S2; Additional file 4: Fig. S4). This is not surprising since Xa1 and its functional homologs exhibit broad-spectrum resistance against African Xoo specifically, owing to the lack of iTALEs in examined African Xoo strains (Ji et al. 2016; Read et al. 2016). The Xa1 allele that we discovered in the rice variety CT13432 appears to contain seven LRRs, like the allele originally identified as Xa45(t), which was cloned from the wild rice variety Oryza nivara-1 (Ji et al. 2020). Other alleles reported to date contain 4 (Xa14), 5 (Xa2 and Xo1), and 6 repeats (Xa1). Variety FKR47N contains an allele that appears to have 6 LRRs. FKR47N, also called NERICA 17, is the result of crossing O. glaberrima variety CG14 and O. sativa ssp. japonica variety WAB181-18, backcrossing to WAB181-18 as recurrent parent. Xa1 alleles have been reported in other NERICAs. Alleles in NERICAs 5 and 7, the recurrent parent of which is the O. sativa ssp. japonica rice variety WAB56-104, have five and six LRRs respectively. NERICAs 12 and 14, coming from the O. sativa ssp. japonica rice variety WAB56-50, each carry an allele with five LRRs (Ji et al. 2020). According to Ji et al. (2020), a search for Xa1 allelic members in the 3000 rice genomes revealed that approximately 15% contain the Xa1 signature sequence. A phenotypic screening for Xa1-like resistance in 87 rice accessions revealed its presence in 16 of them, including different rice species such as O. glaberrima, O. nivara and O. sativa (Ji et al. 2020). A complementary study on more than 500 O. sativa accessions revealed that Xa1 alleles were present in aus, indica, temperate and tropical japonica (Zhang et al. 2020). The frequency of Xa1 alleles among diverse rice varieties and the variability of LRRs number among Xa1 alleles observed in these studies and revealed further by our results, underscores the unanswered question of the origin of the gene and the drivers of its diversification.
BLB represents a serious threat to rice production in Africa. Varietal resistance is the best strategy to control BLB durably, but it requires sources of resistance effective against the local pathogen genotypes. So far, no African strain of Xoo with iTALE/truncTALE has been identified, which makes Xa1 and its alleles promising tools to control BLB in Africa. Although the durability of Xa1 in Africa is difficult to predict, the risk of emergence of strains of Xoo originating from Asia and their spread through the African continent seems high in the current context of intense global exchanges of rice germplasm and insufficient phytosanitary measures. Pyramiding of resistance genes against BLB is of great value to extend the durability and the spectrum of resistance within a rice variety (Oliva et al. 2019). CT13432 combines several rice blast resistance genes (Pi1, Pi2 and Pi33) (Tharreau et al. 2007; Utami et al. 2011), but here we found out that it also carries at least two, distinct types of resistance to BLB, an apparent Xa1 allele and a putative, TalI-dependent E gene, making it an excellent material for breeding to create improved varieties for Africa. What is more, we expect that CT13432 might also carry yet another blast resistance gene, Pi63, which resides at the same locus as Xo1 in Carolina Gold Select (Read et al. 2020b).