Improved bacterial leaf blight disease resistance in the major elite Vietnamese rice cultivar TBR225 via editing of the OsSWEET14 promoter

TBR225 is one of the most popular commercial rice varieties in Northern Vietnam. However, this variety is highly susceptible to bacterial leaf blight (BLB), a disease caused by Xanthomonas oryzae pv. oryzae (Xoo) which can lead to important yield losses. OsSWEET14 belongs to the SWEET gene family that encodes sugar transporters. Together with other Clade III members, it behaves as a susceptibility (S) gene whose induction by Asian Xoo Transcription-Activator-Like Effectors (TALEs) is absolutely necessary for disease. In this study, we sought to introduce BLB resistance in the TBR225 elite variety. First, two Vietnamese Xoo strains were shown to up-regulate OsSWEET14 upon TBR225 infection. To investigate if this induction is connected with disease susceptibility, nine TBR225 mutant lines with mutations in the AvrXa7, PthXo3 or TalF TALEs DNA target sequences of the OsSWEET14 promoter were obtained using the CRISPR/Cas9 editing system. Genotyping analysis of T0 and T1 individuals showed that mutations were stably inherited. None of the examined agronomic traits of three transgene-free T2 edited lines were significantly different from those of wild-type TBR225. Importantly, one of these T2 lines, harboring the largest homozygous 6-bp deletion, displayed decreased OsSWEET14 expression as well as a significantly reduced susceptibility to a Vietnamese Xoo strains and complete resistance to another one. Our findings indicate that CRISPR/Cas9 editing conferred an improved BLB resistance to a Vietnamese commercial elite rice variety.


Introduction
Here, we report on the identification of OsSWEET14 as a transcriptional target of Vietnamese Xoo. CRISPR/Cas9-mediated mutagenesis of the OsSWEET14 promoter in TBR225, a major elite variety in rice production areas of North Vietnam, is shown to confer BLB resistance without detectable yield penalty. The current study found this quintessential S gene to be associated with the virulence of Vietnamese Xoo strains. This is an important step for the future design and implementation of broad-spectrum BLB-resistance in elite rice varieties using genome editing in Vietnam.

Plant and pathogen materials
Rice cultivar TBR225 (Oryza sativa L. ssp. indica) were obtained from ThaiBinh Seed Cor. [27]. All edited and wild-type (WT) TBR225 plants were grown in a net-house under the following average conditions: 30˚C for 14 h (light) and 25˚C for 10 h (dark) with 80% humidity. The Xoo VXO_11 and VXO_15 strains used in this study were isolated from diseased leaves collected in Hanoi-Vietnam in 2013 and 2016, respectively. Bacteria were cultured as described in Zhou et al. (2015) [28].

Gene expression analysis
Gene expression analyses were carried out as described previously [29] by RT-PCR method. The rice leaves were infiltrated with the indicated bacterial strains and used for total RNA extraction 48 h post inoculation using the TRIzol reagent (Invitrogen, USA). One microgram of RNA was used for each RT-PCR with oligo (dT) primer followed by PCR with OsS-WEET14-specific primers (forward 5 0 -ACTTGCAAGCAAGAACAGTAGT-3 0 and reverse 5 0 -AT GTTGCCTAGGAGACCAAAGG-3 0 ). An Eppendorf Mastercycler ep Gradient S was used for 35 PCR cycles. The OsEF1α gene was used as a constitutive control [15] using specific primers (forward 5 0 -GAAGTCTCATCCTACCTGAAGAAG-3 0 and reverse 5 0 -GTCAAGAGCCTCAAGCA AGG-3 0 ).

gRNA design
The OsSWEET14 promoter (GenBank, accession number: AP014967.1) was amplified by PCR with forward primer 5'-TTGCGGCTCATCAGTTTCTC-3' and reverse primer 5'-CTAGGAGAC CAAAGGCGAAG-3' from genomic DNA of TBR225 rice plants and ligated in pGEM-T Easy vector (Promega) for sequencing. The gRNA target sequence ( Fig 1A) for editing the TBR225 OsSWEET14 promoter was designed based on the sequence of the cloned TBR225 OsS-WEET14 promoter using a combination of two bioinformatics tools CRISPR-P v2.0 [30] and CCTop [31]. A gRNA sequence with high on-target and low off-target scores in both prediction tools was chosen for vector construction.

Bacterial blight inoculation
Rice cultivation and disease assays were done according to the methods of Blanvillain-Baufumé et al. (2017) [15]. Bacteria were cultured in PSA media (10 g/liter peptone, 10 g/liter sucrose, 1 g/liter glutamic acid, 15 g/liter Bacto Agar) at 28˚C for two days [35] and inoculated at an optical density (OD 600 ) of 0.5 (infiltrations) or 0.4 (leaf clipping) in water. For lesion length measurements, at least three inoculated leaves per plant and three plants for each line were measured 14 days after inoculation (DAI), and scored as follows: high resistance (lesion length < 8 cm), moderate resistance (lesion length 8-12 cm) and susceptibility (lesion length > 12 cm). For gene expression analyses, 4-cm leaf sections infiltrated with bacterial suspensions were collected at 48 h after inoculation for RNA extraction. Experiments included samples from three pooled biological replicate leaves. The plants inoculated with distilled water only were used as negative controls.

Analysis of OsSWEET14 edited allele sequences
To determine the nature of the mutation at the target site, all transgenic T 0 or T 1 plants were analyzed by PCR using genomic DNA (50 ng) as a template and OsSWEET14 specific primers (5 0 -TTGCGGCTCATCAGTTTCTC-3 0 and 5 0 -TGGATCAGATCAAAGGCAAC -3 0 ). The PCR products were directly sequenced using the Sanger method. The sequencing chromatograms were decoded using the Degenerate Sequence Decoding method [36] in order to identify the mutations.

Evaluation of major agronomic traits under net-house conditions
WT and selected mutant plants were planted under net-house conditions in a randomized pot design experiment. At maturity, five plants of each line were investigated for the following agronomic traits: growth duration, plant height, number of tillers per plant, number of grains per panicle, number of filled grains per panicle and yield (seed mass) per plant. The experiment was repeated three times, so a total of fifteen plants were evaluated for each line.

Analysis of potential off-target editing
Off-target sequences were predicted with the CCTop tool [31] against the OsSWEET14 promoter sgRNA and the rice Nipponbare genome with default parameters. A total of 18 potential off-target sequences were identified. Three of them were located in coding regions (S2 Table). These regions were amplified by PCR using the specific primers listed in S2 Table and analyzed by sequencing.

Vietnamese Xoo strains induce OsSWEET14 during infection of the TBR225 rice variety
OsSWEET14/Os11N3 was previously identified as a susceptibility gene for Xoo strains relying on either of the AvrXa7, PthXo3, TalF (formerly Tal5) or TalC TALEs for infection of the rice cultivars Nipponbare and Kitaake [11]. Because Xoo strains tend to frequently target this gene, we first sequenced a region of the OsSWEET14 promoter from rice cultivar TBR225 to examine if it also carries documented target EBEs. Based on the Nipponbare genome sequence in database (AP014967.1), the region encompassing 1343 bp sequence upstream and 52 bp sequence downstream of the predicted transcription start site of OsSWEET14 gene from TRB225 rice cultivar was PCR amplified and sequenced (S1 Fig). The promoter region including the putative TATA box (TATAAA) and the AvrXa7, PthXo3, TalF/Tal5 and TalC EBEs (Fig 1A), located 319 bp to 216 bp upstream of the ATG initiation codon, showed 100% identity to the Nipponbare sequence. This therefore implied that in principle, the TBR225 OsS-WEET14 promoter can be recognized by characterized major Xoo TALEs.
As shown in Fig 2A, we also challenged TBR225 plants with two Vietnamese Xoo strains VXO_11 and VXO_15, both originating from the Hanoi province, using leaf clipping assays. We consistently obtained typical extended disease lesions 14 days after inoculation (25.5 cm and 26.6 cm average lesions length for VXO_11 and VXO_15, respectively in the experiment of S2 Fig), indicating that the TBR225 variety is susceptible to BLB.
To test if OsSWEET14 is a potential direct virulence target of Vietnamese Xoo strains, we infiltrated TBR225 rice leaves with the two Vietnamese Xoo strains. Forty-eight hours post infiltration, TBR225 plants inoculated with VXO strains displayed a strong induction of OsS-WEET14 relative to water controls ( Fig 2B). These results suggest that OsSWEET14 is a transcriptional target of VXO strains and that it may act as a susceptibility gene in TBR225.

CRISPR/Cas9 design for OsSWEET14 promoter editing
Our main objective was to engineer resistance to BLB caused by Vietnamese Xoo strains. To this end, we subsequently sought to specifically modify the OsSWEET14 promoter in TBR225 rice with CRISPR/Cas9-mediated editing. Previous work revealed that while African Xoo strains rely on TalC and occasionally, TalF, all Asian Xoo strains use either PthXo3-or Avr-Xa7-like TALEs to activate OsSWEET14 [11]. Because the talC gene is currently exclusively found in African strains, we reasoned that it is unlikely that Vietnamese strains carry a talC copy. Thus, to maximize our chances to perturb all remaining documented EBEs, we selected a 20-bp nucleotide target site overlapping the PthXo3, AvrXa7 and TalF EBEs and having a predicted cut site located near the 3'-end of the AvrXa7 EBE ( Fig 1A). The recombinant binary plasmid pCas9/OsSWEET14-gRNA for CRISPR/Cas9 mediated editing of OsSWEET14 was transformed into the rice variety TBR225 via Agrobacterium-mediated transformation (S1 Table). A total of nine TBR225 transformants were selected from 10 independent PCR-validated transgenic T 0 TBR225 plants to further investigate CRISPR/Cas9-targeted mutagenesis of the OsSWEET14 promoter. In order to decipher the nature of the editing events in OsS-WEET14, the promoter sequencing data of transgenic lines were analyzed using the Degenerate Sequence Decoding software [36]. All 9 T 0 transgenic plants harbored at least an editing event ( Fig 1C): two were heterozygous mutant/wild type, two had homozygous mutations, and five had bi-allelic mutations. Regarding the type of mutations, 66.7% were nucleotide deletions, 11.1% of the mutations were nucleotide insertions and no substitution was detected ( Table 1).

Inheritance of CRISPR/Cas9-induced mutations in the T 1 generation
To assess the inheritance of the CRISPR/Cas9-induced OsSWEET14 mutations in the next generation, all T 0 mutant transgenic plants ( Fig 1C) were allowed to self-pollinate, and T 1 transgenic plants were randomly selected in the progeny of T 0 plants for sequencing and analysis of their edited site ( Table 2). All T 1 individuals derived from T 0 plants previously genotyped as homozygous possessed the same allele as their parent, suggesting stable inheritance of the mutations to the next generation. Similarly, the T1 progeny of each of both bi-allelic and heterozygous mutation T 0 lines showed a segregation ratio which is consistent with Mendelian segregation (χ 2 < χ 2 0.05, 2 = 5.99), indicating that the CRISPR/Cas9-induced mutations in T 0 plants were transmitted as expected to the next generation. Interestingly, no new mutant allele was detected in the T 1 generation of both heterozygous mutants L-21 and L-27, even though most of them still carried the transgene. Overall, consistent with previous similar studies, our results indicate that the CRISPR/Cas9-mediated mutations generated here are stably transmitted to the next generation in a Medelian fashion.

Selection of transgene-free mutant TBR225 rice lines
To identify T-DNA free T 1 rice plants containing a mutation in EBEs of the OsSWEET14 promoter, PCR analysis was carried out using primers specific to Cas9, sgRNA and HPT sequences ( Table 2). A T1 individual was considered devoid of the transgene if the control amplification of the OsSWEET14 promoter was successful and if none of the PCR reactions with independent primer pairs designed on the T-DNA produced a detectable diagnostic band. The results of this PCR screen show that the T-DNA could be segregated out in the progeny of most T 0 lines, with 88.9% of the T 0 lines generating T-DNA-free progeny. In total, 44 of 221 analyzed edited T 1 plants did not generate a specific amplicon from the T-DNA construct and 15 of them were homozygous mutant harboring the desired OsSWEET14 modifications. Our results demonstrate that transgene-free, homozygous mutant individuals could be obtained in the segregating progeny of selfed T 0 individuals.

TBR225 OsSWEET14 promoter editing confers resistance to Vietnamese Xoo
To characterize the BLB-resistance phenotype of the generated rice mutants, three T-DNAfree, homozygous TBR225 edited lines, namely, L-5.7(-6), L-31.12(-3) and L-15.4(+1) with OsSWEET14 promoter alleles corresponding respectively to L-5-a1 (6bp deletion), L-31 (3bp deletion) and L-15 (1bp insertion) in Fig 1C, were established. Selected T 1 individuals were propagated to obtain T 2 seeds which were used to perform BLB susceptibility assays. Edited T 2 and WT TBR225 plants were inoculated by leaf-clipping with the VXO_11 and VXO_15 strains at the eight-week stage. The inoculated leaves of wild type TBR225 plants and of edited lines L-15.4(+1) and L-31.12(-3) developed long water-soaked lesions typical of BLB, ranging from 18.3 cm to 29.0 cm in length. In contrast, the edited line L-5.7(-6), harboring a longer 6-bp deletion at the target site, displayed high (1.2 cm average lesion length) and moderate (7.3 cm average lesion length) resistance to VXO_11 and VXO_15 strains, respectively (Fig 3). Means comparisons with a Tukey's HSD test further indicated that irrespective of the inoculated strain, the mean lesion lengths measured on the L-15.4 (+1), L-31.12(-3) or wild type lines were not significantly different. In contrast, the mean lesion lengths recorded on the L-5.7(-6) mutant line were significantly different from those obtained on the wild type and the two other edited lines challenged with either of the Vietnamese strains (Fig 3B). Furthermore, our off-target editing analysis on line L-5.7(-6) did not reveal unintended modifications of other annotated rice loci (S2 Table and S5 Fig), indicating that the 6-bp deletion in the OsSWEET14 promoter is probably responsible for this phenotype. Consistent with disease assays and as shown in Fig 3C, whereas a semiquantitative RT-PCR signal for OsSWEET14 expression was detected on the parental variety and the L-15.4(+1) and L-31.12(-3) edited lines following VXO_11 and VXO_15 infiltration, this amplicon was undetectable in the resistant L-5.7(-6) line.
In conclusion, this data shows that the 6-bp deletion in the AvrXa7/PthXo3 EBE reduces dramatically OsSWEET14 expression following VXO strains inoculation and confers resistance to these strains. In contrast, shorter modifications on the 3'-end of this EBE are insufficient to perturb OsSWEET14 expression after inoculation and do not confer detectable protection against the corresponding strains. Finally, while these results strongly support the view that OsSWEET14 functions as a unique susceptibility gene in the interaction between strain VXO_11 and the TBR225 rice variety, the resistance to strain VXO_15 is not as dramatic and may suggest that other mechanisms partially counteract the effects of the AvrXa7/PthXo3 EBE 6-bp deletion in edited TBR225 plants.

TBR225 OsSWEET14 promoter edited lines agronomic performances are undistinguishable from the parental variety
To determine if mutations in the OsSWEET14 promoter affect agronomic traits of TRB225 rice plants, three independent homozygous mutant lines were analyzed by measuring their growth duration, plant height, number of tillers per plant, number of grains per panicle, number of filled grains per panicle, yield per plant and amylose content under net-house conditions (see picture of S3 Fig). ANOVA tests and Student's t tests showed that the mutant lines displayed no significant difference to TBR225, in terms of the examined agronomic traits, under our net-house conditions (Table 3). These results suggest that the tested CRISPR/Cas9-induced mutations in the OsSWEET14 promoter did not negatively impact the main agronomic traits of TBR225.

Discussion
Recently, the CRISPR/Cas9 system has emerged as a powerful tool for gene editing in many organisms including plants. Because of its specificity and efficiency, this system has been widely used to improve important agronomic traits of major crops such as rape, tomato, soybean, rice, wheat and maize [37]. Excluding easy-to-transform reference accessions such as Nipponbare and Kitaake that are widely used in the laboratory, the number of reports on the
TBR225 [27], a major commercial rice variety cultivated in large areas of Northern Vietnam, has the advantages of early maturity, high and stable yield, as well as cooking quality. However, it is very susceptible to BLB. Here, the CRISPR/Cas9-mediated editing method was applied in order to rapidly improve the BLB resistance of TBR225 by modifying the AvrXa7, PthXo3 and TalF EBEs on the promoter of OsSWEET14. Of the three generated homozygous mutant lines tested for resistance, the one carrying the largest deletion at the target site (6 bp) showed a significantly improved resistance to infection with two Xoo strains VXO_11 and VXO_15. Therefore, using the major commercial rice variety TBR225 as an example, we illustrate the advantages of CRISPR/Cas9 tool for rice breeding.
In the present study, the frequency of individuals with CRISPR/Cas9-induced mutations in T 0 transgenic plants was 90%, which is similar to previous observation [33]. We obtained only two heterozygous mutant/wild type lines versus seven homozygous or bi-allelic mutant lines. This high frequency of mutated alleles is another proof that the CRISPR/Cas9 system is indeed an efficient tool for gene editing in plant. We also observed the stable transmission of edited alleles to subsequent generations. This is a common phenomenon that has been repeatedly documented for rice plants carrying CRISPR/Cas9-induced mutations [38,40,43,44]. In this study, we obtained only two types of induced mutations in T 0 plants: insertion (11.1%) and deletion (66.7%), but no substitution were observed. In some earlier studies, new mutations were continuously obtained in the T 1 offspring of heterozygous T 0 mutants because the Cas9 complex remains active on edited targets until the seed or PAM regions cease to be functional [35,37,43]. In contrast, here, all the T 1 plants generated from both heterozygous lines L-21 and L-29, regardless of whether they had a CRISPR/Cas9 T-DNA transgene integrated in their genome, did not show any new mutation. We could also readily obtain transgene-free plants from most of the T 1 segregation populations without any laborious crossing or backcrossing steps, which illustrates an advantage of the CRISPR/Cas9 technology compared to conventional breeding.
Clade III SWEET family proteins are involved in a number of biological processes such as seed and pollen development or pathogen susceptibility [45]. Their inactivation has previously been shown to cause pleiotropic and/or detrimental effects. For example, both ossweet11 single and ossweet11-ossweet15 double Kitaake rice mutants showed defects in endosperm development and filling [46]. In addition, RNA-mediated silencing of either Os11N3/OsSWEET14 [7] or Os8N3/OsSWEET11 [6] in BLB resistant Kitaake lines causes negative effects on seed production. In contrast, here, we show that T-DNA-free TBR225 plants harboring homozygous mutations generated with the CRISPR/Cas9 system in the AvrXa7/PthXo3 EBE of the OsS-WEET14 promoter exhibited enhanced Xoo resistance but did not show any significant difference in all examined agronomic traits compared to wild-type plants under net-house growth conditions. It is conceivable that limited modifications in promoter regions do not affect the normal expression of SWEET genes in contrast to KO or silenced lines. Our findings are consistent with the previous work of Oliva et al. [11] who studied 30 combinations of EBE mutations in the OsSWEET11, OsSWEET13 and OsSWEET14 promoters of the IR64 or Ciherang-Sub1 varieties and detected only a single line with abnormal agronomic traits. Some individual Xoo strains have evolved a set of distinct TALE effectors that collectively target several members of the clade III SWEET family. The presence of these redundant TALEs thereby trumps single "loss-of-tale-responsiveness" resistance alleles [11,12,17,47]. For example, Kitaake lines carrying TALEN-induced mutation in the SWEET14 promoter [13,15] exhibit resistance to strains which depend exclusively on matching AvrXa7/PthXo3 for clade III SWEET family induction. Likewise, the natural xa13 allele [48] or CRISPR/Cas9-induced mutation in the SWEET11 promoter [11] exhibit resistance to strains such as PXO99 which depend exclusively on PthXo1, for virulence. However, the BLB resistance of the Kitaake lines harboring mutations in both AvrXa7/PthXo3 (OsSWEET14) and PthXo1 (OsSWEET11) EBEs was defeated by Xoo strains expressing simultaneously the AvrXa7/PthXo3 and PthXo2B TALEs [11]. Recently, the stacking of EBE-edited alleles in several OsSWEET promoters have overcome this limitation and was shown to achieve a broad spectrum of resistance to strains from most BLB-prone countries in Asia [11,12].
All of the three T 2 lines tested for BLB resistance were affected for the AvrXa7/PthXo3 EBE and conserved an otherwise wild type TalF EBE (Fig 1C). The homozygous mutant TBR225 line L-5.7(-6) carrying a 6-bp deletion in the AvrXa7/PthXo3 EBE exhibited a significantly enhanced resistance to two Vietnamese Xoo strains compared to WT TBR225. The L-15.4(+1) and L-31.12(-3) lines that harbored more subtle alterations in the 3'-end of this EBE (a 1-bp insertion and a 3-bp deletion, respectively) in contrast remained susceptible to VXO strains. Our OsSWEET14 expression analysis after Vietnamese Xoo strains inoculation (Fig 1C) suggests that these editing events did not alter the EBE sequence sufficiently to compromise promoter recognition by an AvrXa7/PthXo3-like Vietnamese TALE. With less than 2 cm average lesion length, the resistance of line L-5.7(-6) (6-bp deletion) to the VXO_11 strain is rather extreme (versus average lesion length of 20.1 cm on wild type plants). Moreover, in this line, OsSWEET14 expression following bacterial inoculation is strongly reduced relative the parental line and the two other edited lines, which suggest that in this case, recognition by an AvrXa7/PthXo3-like Vietnamese TALE is abrogated. Consistent with OsSWEET14 expression analysis and as shown in S4 Fig, the Talvez [49] target prediction scores for AvrXa7 and PthXo3 on the OsSWEET14 promoter L-5-a1 allele sequence of line L-5.7(-6) are markedly lower than on the wild type promoter sequence. This is not the case however for the edited alleles carried by lines L-15.4(+1) and L-31.12(-3) (respectively L-15 and L-31 in S4 Fig) whose Talvez scores are identical or slightly lower than those of the wild type promoter sequence.
The magnitude of the effect of the 6-bp deletion allele on susceptibility to VXO_11 is comparable to the dramatic effect of previously characterized alterations of the same EBEs in the Kitaake background against the PXO86 strain that possesses a single TALE, AvrXa7, targeting OsSWEET14 for clade III OsSWEET gene induction [15]. By analogy, this suggests that OsS-WEET14 is also the only clade III OsSWEETs target of VXO_11 in the TBR225 background but, in order to confirm this hypothesis an examination of other clade III OsSWEET genes expression patterns in response to this strain would be required. The situation with the VXO_15 strain is not as straightforward to interpret and will require further investigations. Although the 6-bp deletion in the AvrXa7/PthXo3 EBE did provide an increased resistance to the edited plants, the VXO_15 strain caused intermediate disease severity (7.3 cm average lesion length on Fig 3). This incomplete resistance is unlikely to result from the partial but still productive recognition of subsequences of the altered EBE by a VXO_15 AvrXa7/PthXo3-like TALE because OsSWEET14 expression is similarly decreased in response to either this strain or VXO_11 (Fig 3C). Alternatively, contrary to all Asian Xoo examined so far, but similar to African Xoo [11,15], VXO_15 may have the intrinsic potential to cause disease in the absence of clade III OsSWEET gene induction, a phenomenon that seems to be dependent on the edited rice variety genetic background [42]. More likely, analogous to other Asian strains, VXO_15 may encode alternative TALEs, such as PthXo2B or PthXo1 that compensate the loss of OsSWEET14 induction by targeting other clade III OsSWEET genes. In this regard, deciphering clade III OsSWEET genes expression patterns in combination with long read genome sequencing will ultimately help describe TALEs variability in Vietnamese Xoo strains and its functional impact on OsSWEET genes induction.
In conclusion, we showed that editing specific EBEs of Xoo TALEs via CRISPR/Cas9 tool is an efficient method for improving BLB resistance of elite rice varieties such as TBR225 without detectable yield penalties. This also uncovered the potential diversity of TALEs in Vietnamese Xoo population, which will thus require future investigations to address the TALE repertoires of Vietnamese Xoo strains in order to generate broad-spectrum BLB-resistant rice varieties in Vietnam.  Fig 3C for