Mutations introduced in susceptibility genes through CRISPR/Cas9 genome editing confer increased late blight resistance in potatoes

The use of pathogen-resistant cultivars is expected to increase yield and decrease fungicide use in agriculture. However, in potato breeding, increased resistance obtained via resistance genes (R-genes) is hampered because R-gene(s) are often specific for a pathogen race and can be quickly overcome by the evolution of the pathogen. In parallel, susceptibility genes (S-genes) are important for pathogenesis, and loss of S-gene function confers increased resistance in several plants, such as rice, wheat, citrus and tomatoes. In this article, we present the mutation and screening of seven putative S-genes in potatoes, including two DMR6 potato homologues. Using a CRISPR/Cas9 system, which conferred co-expression of two guide RNAs, tetra-allelic deletion mutants were generated and resistance against late blight was assayed in the plants. Functional knockouts of StDND1, StCHL1, and DMG400000582 (StDMR6-1) generated potatoes with increased resistance against late blight. Plants mutated in StDND1 showed pleiotropic effects, whereas StDMR6-1 and StCHL1 mutated plants did not exhibit any growth phenotype, making them good candidates for further agricultural studies. Additionally, we showed that DMG401026923 (here denoted StDMR6-2) knockout mutants did not demonstrate any increased late blight resistance, but exhibited a growth phenotype, indicating that StDMR6-1 and StDMR6-2 have different functions. To the best of our knowledge, this is the first report on the mutation and screening of putative S-genes in potatoes, including two DMR6 potato homologues.

were maintained in vitro by sub-culturing the apical portion of 3-4 week-old stems on Murashige and Skoog (MS) basal nutrient including vitamins (Duchefa, M0222.0050) with 10 g/L sucrose and 7.5 g/L Phyto agar (MS10) 24 . Genetically modified lines containing three resistance genes, 3R, Rpi-blb2, Rpi-blb1, and Rpi-vnt. 1 7,24 , in Désirée and King Edward were used as resistant controls. The P. infestans strain 88,069 (A1 mating type, race 1.3.4.7) was propagated as previously described 25 . Vector constructs. Candidate genes were selected (Table 1) and the coding sequence analysed for possible CRISPR targets and their number of off-targets using Cas-designer (http://www.rgeno me.net/cas-desig ner); 26 and CRISPOR (https ://crisp or.org); 27 . For each candidate, two PCR primer pairs were designed to amplify a region containing putative targets with the fewest potential off-targets and used in PCR amplification of genomic DNA and cDNA (see Supplementary Table). PCR products were run on 1% agarose gels, gel-purified, and each band was sequenced using two primers. For each candidate, the two targets that were conserved in all sequences, and that had the lowest number of potential off-targets were selected (see supplementary Fig. 1). The targets were assembled into the Csy4 multi-gRNA vector pDIRECT_22C, using protocol 3A 22 to form the plasmid pDIRECT_22C_S-gene.
Potato transformation protocol. The protocol for the Agrobacterium transformation of S. tuberosum Désirée and King Edward was modified from the original protocol 24,28 . A 10 mL overnight liquid culture of Agrobacterium tumefaciens C58 carrying the plasmid of interest was centrifuged at 5000 rpm in a 15 ml tube for 10 min, the supernatant was discarded, and the pellet was re-suspended in 10 mL dH 2 O containing 50 µl of acetosyringone (76 mM). For transformation, 1 mL of the Agrobacterium suspension (OD 1.9-2.0) was pipetted onto dissected leaf explants that were placed on the co-cultivation media. Leaf explants were incubated under reduced light (50% intensity) for 48 h before they were transferred to selective media (400 mg/L cefotaxime + 100 mg/L kanamycin, and 2 mg/L for Désirée and 5 mg/L for King Edward of zeatin ribose) for regeneration 24 . Leaf explants were sub-cultured onto fresh media every 7-10 d to maintain selection pressure. Shoots that emerged after 4-5 weeks were dissected and rooted on MS media containing no plant growth regulators but with continued selection (100 mg/L kanamycin). Only shoots that initiated roots in the selective media were screened at the molecular level.
PCR screening and sequencing. Genomic (see table S1), and 30 s at 72 °C, with a final extension at 72 °C for 5 min. The samples were analysed on 2% agarose gels (except the CHL gene, 3% agarose gels were used) and tetra-allelic deletion mutant lines were selected (except the HDS gene, see results). Each PCR band was isolated from agarose gels and purified using a GeneJET Gel Extraction Kit (Thermo Fisher Scientific, Waltham, USA). Purified samples were sequenced at Eurofins Genomics (Germany), see supplementary figure S2. Detached-leaf assay. For each experiment, nine fully developed leaves from 5-week-old plants from each line were used for detached-leaf assays (DLAs). The inoculum of P. infestans was prepared by harvesting sporangia from 12 to 14 d-old plates of P. infestans in clean tap water 32 . The inoculum was adjusted to 20,000 sporangia/mL and 25 µL of the spore solution was pipetted onto the abaxial side of the leaflet. The infected leaves were maintained in a humid environment (RH ~ 100%) under controlled conditions 33 . Results were recorded by measuring the infection size of each leaflet at 7 d post-inoculation (dpi). The difference between the means was tested using a t-test with the significance level of p < 0.05 or 0.01. We also calculated the percentage of successful infection.

Result and discussion
Selection of putative S-genes in Potato against Phytophthora infestans. S-genes involved in susceptibility to different types of pathogens have been found in many different plant species 17,34 . Here, S-gene candidates were selected based on the following criteria: pathogen resistance phenotype, being either a single gene or belonging to a small confined gene family in potatoes, each S-gene concerning other candidates should have a different function, and if possible, function in different pathways (see Table 1). MLO (Mildew resistance locus) encodes a plasma membrane-localized seven transmembrane domain protein associated with vesical transport and callose deposition 8,9,35 . The MLO protein contains a domain that is predicted to bind with calmodulin and is required for full susceptibility to powdery mildew infection 9 . In this study, we included MLO because it is a typical S-gene, which has been successfully applied in many plants, such as roses, peas, melons, and apples 9 . Furthermore, mlo mutants also showed resistance to two oomycetes: the hemibiotrophic Phytophthora palmivora 10 and the biotrophic Hyaloperonospora arabidopsidis 36 . Because P. infestans also is a oomycete with a hemibiotrophic lifestyle, we decided to include this gene in the screening. Appiano et al. (2015) identified the corresponding MLO gene in potatoes and named it StMLO1 37 .
In Arabidopsis, HDS encodes a chloroplast localized hydroxy-2-methyl-2-(E)-butenyl 4-diphosphate synthase, one of the last steps in the methylerythritol 4-phosphate (MEP) pathway from which chlorophyll, carotenoids, gibberellins, and other isoprenoids are derived 38 . HDS is a negative regulator of salicylic acid (SA) by reducing the amount of its substrate, methylerythritol cyclodiphosphate (MEcPP) 46 . Arabidopsis HDS mutant plants show enhanced resistance to biotrophic, but not to necrotrophic, pathogens 47 . In potatoes, we only encountered one HDS gene homologue.
The triphosphate tunnel metalloenzymes (TTMs) hydrolyse organophosphate substrates 39 . Arabidopsis encodes three TTM proteins, where TTM2 is involved in pathogen resistance via an enhanced hypersensitive response and elevated SA levels 48 . Atttm2 mutant lines showed enhanced resistance to the biotrophic pathogen Hyaloperonospora arabidopsidis. The closest potato homologues to the AtTTM2 gene are DMG400025117 and DMG400001931. DMG400025117 appeared to be induced by the SA homologue BTH, whereas DMG400001931 was not (http://bar.utoro nto.ca/efp_potat o/cgi-bin/efpWe b.cgi); therefore, we chose to analyse DMG400025117 since late blight resistance is influenced by SA. Furthermore, as TTM2 has only been studied in Arabidopsis, its relevance in acquiring resistance in crop plants is unknown. Sun et al. (2016Sun et al. ( , 2017 analysed potato plants, where StDND1 had been knocked-down using RNAi and found that the plants were more resistant toward P. infestans. StDND1-silenced plants displayed auto-necrotic spots only in the leaves of older plants and a few well-silenced StDND1-transformants showed dwarfing 12 , a phenotype that might result from inadequate specificity of the RNAi approach or the efficiency of silencing may fluctuate during development. The DND1 gene encodes a cyclic nucleotide-gated ion channel, which has been implicated in Ca 2+ signalling related to various physiological processes (pathogen defence, development, and thermotolerance) 49 .
StCHL1 is a putative S-gene in potatoes. Originally, StCHL1 was found through microarray analysis of brassinosteroid responsive marker genes in potatoes. Gene overexpression and virus-induced gene silencing experiments showed this gene to be important for P. infestans colonization of Nicotiana benthamiana 42 . No experiments in potato has been carried out. CHL1 is a transcription factor, which regulates brassinosteroid hormone signalling and immune response 50 ; in potatoes, we located only one such gene.
DMR6 proteins belong to the 2-oxoglutarate (2OG)-Fe (II) oxygenase family. In Arabidopsis, AtDMR6 encodes an SA 5-hydroxylase that regulates SA homeostasis by converting SA to 2,5-DHBA 45 . This gene is a negative regulator of the active SA pool; thus, it is important for the SA-dependent plant immune system. Knockout of SlDMR6-1 in tomatoes enhanced the resistance to Phytophthora capsici and Pseudomonas syringae 43 . Two DMR6 homologues were identified in potatoes. Knockdown of StDMR6 in potatoes by RNAi showed an unclear resistance phenotype, with only six out of 12 transformed plants showing lower transcript levels of DMR6 and four plants showed a resistance phenotype, whereas eight plants showed susceptibility to Phytophthora infestans 12 . Therefore, both potato DMR6 homologues were investigated separately by knockout experiments with genome editing.
Efficiency of double guide mediated tetra-allelic mutation varied between genes. By applying two guide RNAs, targeted deletions in the gene of interest may be generated 22,23 . In a study by Čermák el al. 2017, deletions between the two cleavage sites were far more prevalent than individual indels resulting from cleavage of a single site. Therefore, we used the pDIRECT_22C vector 22 encoding two guide RNAs for knocking out S-genes in potatoes. For our screen of edited potato plants, we chose to use PCR with gene-specific primers, spanning both gRNA targets, followed by gel electrophoresis analysis, as a simple, inexpensive, and rapid method for detecting deletions in the target gene. The screening results are shown in Fig. 1 for the lines that were subsequently screened for late blight resistance and growth phenotypes. Sequence data of the target regions is shown in supplementary figure S2.
The number of plants with a deletion in all four alleles was related to locus and target sequence ( Table 2). Analysis of shoots showed variation in the prevalence of tetra-allelic deletion mutants ranging from 0 to 18%. This number can be regarded as the minimum number because we did not detect single nucleotide mutations with this PCR method, but because it was easy to generate many lines in potatoes we believe this was the most efficient method. Analysing in silico target efficiency with several different online tools did not reveal a specific tool that could predict the mutation rate better than others (Table 2).
In Arabidopsis, homozygous mutation of HDS caused an albino phenotype and seedling lethality 38 . In the present study, in agreement with this observation, some calli turned white and did not develop into seedlings. Furthermore, none of the StHDS genome-edited seedlings were confirmed to be deleted in all four alleles.  www.nature.com/scientificreports/ Therefore we concluded that, as in Arabidopsis, a full tetra-allelic HDS deletion is lethal, although transformed cells with a mutation in one, two, or three alleles were able to develop and form shoots ( Table 2). For all other genes, full allelic knockouts were not linked with lethality. Two genes showed a high number of tetra-allelic deletion mutants, namely 13% of StMLO1 and 18% of StCHL1 shoots had a deletion in all four alleles. The other four genes showed a prevalence of between 0.7% and 2.4% tetra-allelic deletion mutants. As mentioned above, because the applied PCR screening did not detect point mutations or very short deletions/insertions, the number of mutants detected in the present study may be lower than that of other screening methods, such as CAPS (Cleaved-Amplified-Polymorphic-Sequence) or IDAA 19 . However, a combination of constructs expressing two gRNAs with PCR screening of shoots is a low-cost, simple, and fast method enabling large scale screening at the shoot level (Fig. 1, supplementary Fig. 3).

StDND1, StCHL1, and StDMR6-1 tetra-allelic deletion mutants showed enhanced late blight resistance.
To analyse late blight resistance in tetra-allelic mutant lines, DLAs were performed. Infection lesion diameter was determined 7 days after P. infestans inoculation (Fig. 1) and the percentage of infected leaves was analysed ( Table 3).
Knockout of StMLO1 in potatoes did not increase late blight resistance as evident by the sizes of the lesion or percentage of infected leaves. Nor there any growth phenotype was detected ( Fig. 2A). The effect on P. infestans infection in mlo potatoes was tested in the present study for the first time. All eight Stmlo1 mutant lines were Table 2. Summary of screening of deletion mutants in this study.  Table 3. Percent of successfully infected leaflets in detached-leaf assay. Mut-1 to Mut-8 are mutant lines and correspond to the lines in Fig. 1 (from left to right). Leaflets from 5-week-old plants were inoculated with 25 μL 20,000 sporangia/mL. Results were scored 7 dpi and a total of nine leaflets per line were used.  (Fig. 1A, Table 3). This was somewhat unexpected because the mutation of orthologous MLO genes is effective in many plant and pathogen species 36,37 , including the hemibiotrophic P. palmivora. Silencing of Capsicum annum CaMLO2 conferred enhanced resistance against virulent Xanthomonas campestris, whereas overexpression of CaMLO2 in Arabidopsis conferred enhanced susceptibility to both Pseudomonas syringae and Hyaloperonospora arabidopsidis 36 . Recently, a wheat mlo mutant was shown to be susceptible to the hemibiotrophic fungal pathogen Magnaporthe oryzae, whereas it was still resistant to the obligate biotrophic fungus Blumeria graminis 11 . Thus, the usefulness of MLO is dependent on the host as well as the pathogen. After PCR screening of 169 putative HDS shoots, we did not obtain any tetra-allelic mutant lines (Table 2). After 2 weeks in soil, some heterozygous mutants showed an albino phenotype (Fig. 2B) and did not grow further, www.nature.com/scientificreports/ whereas shoots with green leaves grew into adult plants. In A. thaliana, the Athds was mutagenized with ethyl methanesulfonate (EMS) and influenced chloroplast development and increased resistance to Pseudomonas syringe 47 . Our potato Sthds mutants showed weakened growth (Fig. 2B) and P. infestans screening of eight mutant lines did not show increased resistance to late blight disease (Fig. 1, Table 3). For StTTM2 (DMG400025117), we analysed five tetra-allelic deletion mutant lines. No mutant line showed any altered phenotype (growth, morphology, or pathogen resistance) when compared with wild-type plants (Figs. 1C, 2C). Analysing TTM2 sequences in Solanum tuberosum, two different StTTM2 genes were identified (DMG400025117 and DMG400001931). The study of Ung et al. (2017) suggested that AtTTM1 and AtTTM2 could functionally complement each other; thus, it is plausible that these genes could be functionally complementary to each other and that a double mutant would show resistance to P. infestans in potatoes. Sun et al. (2016 and used RNAi to knockdown potato StDND1 and found that these plants were more resistant to P. infestans. However, the plants were smaller and showed early senescence and necrotic spots on leaves of older plants. In line with their results, our data showed that the size of infection lesions was strongly reduced in all Stdnd1 mutant lines, whereas the percentage of successful infections was reduced in some of the tetra-allelic lines (Fig. 1C and Table 3). Two mutant lines with wild type and mutant PCR-bands (DND 44, DND 82) showed auto-necrotic spots and late blight resistance in older, but not young leaves ( Figure S4B and S4C).

Gene\line WT Mut-1 Mut-2 Mut-3 Mut-4 Mut-5 Mut-6 Mut-7 Mut-8 3R
The tetra-allelic Stdnd1 mutated potato not only exhibited a late blight resistance phenotype (Fig. 1D) as observed from the results of the earlier RNAi study but also showed pleiotropic phenotypes, such as line DND 583 (Fig. 2D). The tetra-allelic Stdnd1 mutant lines, except for the strong resistance phenotype, also showed reduced growth, long and thin stems, as well as necrosis of all leaves ( Figure S4A). These latter pleiotropic phenotypes were not found in StDND1 RNAi lines 12 maybe because of incomplete silencing. The phenotypes of some of our Stdnd1 mutants (DND 44 and DND 82) and StDND1 RNAi lines were very similar ( Figure S4 and Fig. 3C of Sun et al. 2016). In summary, our results indicated that StDND1, due to the pleiotropic phenotypes observed in the Stdnd1 edited lines, was not a good candidate for application in agriculture.
Stchl1 mutations did not affect morphology or growth phenotype (Fig. 2E). Tetra-allelic mutant plants showed a significant late blight resistance phenotype with reduced lesion sizes (Fig. 1E), but no difference in the percentage of infected leaves (Table 3). This could indicate that the importance of this protein is at the disease developmental stage and not in the initial phase. With a function as a Phytophthora effector target and transcription factor, and being involved in brassinosteroid hormone signalling and immune response to P. infestans 50 , StCHL1 has clear potential as an useful S-gene; possibly when combined with other S-or R-factors to improve pathogen resistance.
CRISPR/Cas9 was applied to knockdown both StDMR6-1 and StDMR6-2, respectively. Tetra-allelic CRISPR/ Cas9 knockdown of StDMR6-1 showed a significant increase in resistance against P. infestans both as measured by infected lesion size and the percentage of infected leaves (Fig. 1F, Table 3). This is in contrast to that of Stdnd1 and Stchl1 knockout plants, which only showed reduced infection lesion sizes ( Fig. 1 and Table 3), but no reduction in the percentage of infected leaves. In tomatoes, the CRISPR-Cas9 mediated mutation of the StDMR6-1 ortholog SlDMR6-1 showed increased resistance to P. capsici and P. syringae pv. tomato 43 , indicating broadspectrum disease resistance function of DMR6-1. In potatoes, knockdown of StDMR6 by RNAi increased late blight resistance without any documented effect on growth phenotype 12 . However, only 33% of the RNAi lines showed an increased resistance phenotype 12 . Tomatoes and potatoes each contain two DMR6 genes ( 43 , Table 1). StDMR6-2 and StDMR6-1 transcripts are approximately 80% identical at the nucleotide level. Because these genes are remarkably similar, RNAi may downregulate both, and therefore knock out of either gene by CRISPR-Cas9 is important for the elucidation of individual gene function.
Genome editing of StDMR6-2 showed that this gene was not involved in susceptibility to P. infestans ( Fig. 1G and Table 3). Five tetra-allelic mutants in two potato backgrounds (Désirée and King Edward) showed the same infection lesion size and percentage of infected leaves as that of the wild type. De Toledo Thomazella et al. (2016) did not study tomato SlDMR6-2 further because of the low expression during pathogen infection.
In conclusion, when comparing the DLA results of mutant lines with both wild type (Désirée and King Edward) and an R-gene containing a transgenic line (3R), we identified three genes (StDND1, StCHL1, and StDMR6-1) that when mutated, increased late blight resistance, whereas mutations in StMLO1, StHDS, StTTM2, and StDMR6-2 did not affect late blight resistance in potatoes.

DMR6-1 mutants had no obvious growth-related phenotypes.
StDMR6-1 is a promising S-gene because tetra-allelic mutants not only showed increased late blight resistance ( Fig. 1F and Table 3) but also did not differ in over-all growth phenotype compared with the wild type (Fig. 2F). Measurement of plant height (Fig. 3A), fresh weight (Fig. 3B) and tuber morphology (Fig. 3E) showed no differences between mutants and wild types. Plants mutated in the orthologous gene SlDMR6-1 in tomatoes, showed disease resistance without any documented effects in growth and development under greenhouse conditions 43 . Therefore, StDMR6-1 may be used in potato breeding to create new potato cultivars with broad-spectrum disease resistance.
StDMR6-2 affect growth phenotypes in potato. StDMR6-1 and its ortholog SlDMR6-1 are important in pathogen susceptibility (Fig. 1) 43 without any obvious growth phenotype (Fig. 3). We investigate the effect of the genome editing of StDMR6-2 on potato phenotype (Figs. 2G,H and 3). Our results did not show any changes in late blight resistance. Analysis of growth phenotype showed that tetra-allelic mutants of StDMR6-2 had significantly lower plant height (Fig. 3C) and fresh weight (Fig. 3D) in both cultivar backgrounds. The plants had the same number of leaves as did the wild type, but their internodes were shorter (Fig. 2G). Furthermore, the tuber eyes of StDMR6-2 mutants did not have the reddish colour (anthocyanin) that is typical of King Edward (Fig. 3F)  www.nature.com/scientificreports/ to allow for structure prediction/comparison, which could shed light on potential substrate/functionality differences between StDMR6-1 and StDMR6-2.

Conclusion
Using CRISPR-Cas9 mediated loss of gene function of seven putative S-genes, we showed that three putative S-genes (StDND1, StCHL1, and StDMR6-1) were involved in late blight susceptibility. Among these three, StDMR6-1 and StCHL1 emerged as promising S-gene targets for the breeding of new disease resistance cultivars because they did not show any growth related phenotype. We also concluded that the pDIRECT_22C vector and the applied deletion screening system expressing two gRNAs for fast PCR mediated screening of full or partial allele knockout was highly efficient and applicable in potatoes. We have produced gene-edited material in popular cultivars that are ready for further tests in field trials. License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.