Three potato genotypes, namely, Kufri Pukhraj (KP), Kufri Jyoti (KJ), and Kufri Chandramukhi (KC), were procured. All the three varieties were grown under an institutional farm field at CSIR-IHBT (latitude 32.0934°N; longitude 76.5439°E), Palampur, India. Among the three varieties used in the experiments, KJ and KP have been previously released as blight disease-tolerant varieties, while KC is considered more susceptible to blight diseases of potatoes (https://cpri.icar.gov.in/AddLink/getIMG/6068). The A. solani (MTCC-10690; accession no. HM484353.1) culture was procured from IMTECH, Chandigarh, India, maintained in a laboratory at 25°C, and used for infection purposes. The fungal culture was inoculated to 1-month-old susceptible potato plants with 100 ml of fungal suspension (2 × 10⁵ conidia ml⁻¹) and sprayed to retain the virulence. Conidia were counted using hemocytometer under a microscope. Plants were kept at a relative humidity of 70% at 26°C with 12-h day/night condition. Fungal mycelia obtained from infected portion using single spore culture methodology (Choi et al., 1999) were used for further maintenance of culture along with subsequent infection work. The identity of fungus being A. solani was further confirmed with the use of Internal Transcribed Spacer (ITS) sequencing using universal primer pairs ITS1–ITS4. The 556-bp ITS sequence is shown in Supplementary Figure S2A. A. solani biomass was measured for infected plants of each variety using fungal-specific primer pairs (AS1 and AS2) by qRT-PCR. The methodology of biomass measurement was followed as described by Kumar et al. (2013).

Total proteome (39,031) was downloaded from the PGSC database and used for the screening of TIR-NBS domain family (PGSC_DM_v3.4_pep_representative.fasta.zip://potatogenomics.plantbiology.msu.edu/index.html). A set of putative TIR-NBS proteins was identified from the complete potato proteome using hidden Markov model (HMM) profiles as previously described (Arya et al., 2014). To improve the numbers of TNL proteins left out by HMM application, another strategy was applied that uses pfam ids and conserved domain (CD) information using the CD database (CDD) of the National Center for Biotechnology Information (NCBI) (Supplementary Figure S1). All the proteins obtained were also cross-checked using MyHits (Pagni et al., 2004) and ScanProsite (de Castro et al., 2006) along with manual curation to ensure the presence of all the three TNL domains. Each transcript was further assessed for LRR motifs manually, using LxxLxLxxN/CxL or LxxLxL consensus (Enkhbayar et al., 2004), where x denoted any amino acid and L signifies leucine, along with LRR pfam id-0562. To rule out the presence of CC domain, MARCOIL HMM with a threshold of 90 was used (Delorenzi and Speed, 2002). “PGSC_DM_v3.4_g2t2c2p2func.txt” file was used to retrieve information regarding TNLs encoding desired transcripts. To examine the structural motifs among the identified TNLs, the predicted protein sequences were submitted for the motif analysis by MEME version 4.9.0 (Bailey et al., 2009).

The scaffold for all twelve chromosomes was obtained from the PGSC database and used for mapping TNL genes. The General Feature Format (GFF) file (PGSC_DM_v3.4_gene.gff) was also retrieved. These TNL genes were graphically portrayed on the chromosome using “PhenoGram” tool of Ritchie Lab (http://visualization.ritchielab.psu.edu/phenograms/plot). To check possible duplicated TNLs in potatoes, the MCScanX program (Wang et al., 2012) was applied with all-vs-all blastp using parameters B = 100, V = 10, E-value 1e−10, and tabular output format.

A phylogenetic tree was prepared using the longest peptide of each TNL gene. The TNL protein sequences were aligned using Clustal W incorporated in MEGA v.7.0 (http://www.megasoftware.net), and a phylogenetic tree was prepared using the neighbor-joining (NJ) method with a bootstrap of 1,000 iterations (Zhang et al., 2018).

About 100 ml of fungal suspension (2 × 10⁵ conidia ml⁻¹) was sprayed on plants. The concentration of conidia and mycelium pieces was calculated using a hemocytometer (Sigma-Aldrich Corp., St. Louis, MO, USA) under a microscope. Sterile water (Mock)-treated plants were used as control. All the plants were kept in a growth chamber with a relative humidity of 70% and temperature of 26°C with 12-h day/night conditions. Leaf samples from each potato cultivars were collected at five time points, viz., 1, 2, 3, 5, and 7 days post inoculation (dpi), from three biological replicates of pathogen-inoculated (treatment) and mock-inoculated (control) plants. Total RNA was isolated as previously described (Ghawana et al., 2011) from leaf tissues collected at different time points. The RNA was reverse transcribed using High Capacity cDNA synthesis kit (Applied Biosystems, Foster City, CA, USA). The cDNA was diluted 10-fold with diethyl pyrocarbonate (DEPC)-treated water. Primer Express software version 3.0.1 (Thermo Fisher Scientific, Waltham, MA, USA) was used to design the primers. The primers were designed with an amplicon size range of 80–190 bp, primer length of 18–25 nucleotides, Tm of 50–60°C, and maximum guanine–cytosine (GC) content of 50%–60%. All primer sequences are presented in Supplementary Table S1. The Power Sybr Green PCR master mix (ABI) was utilized for the gene expression using an optical 96-well plate on MX3000P real-time PCR machine (Agilent Technologies, Waldbronn, Germany) as per the manufacturer’s manual. Elongation factor (StEf1a) (accession no. PGSC0003DMG400023270, XM_006343393.2) was selected as reference gene to normalize the targeted gene expression (Tang et al., 2017). The relative expression was calculated by the 2^−∆∆CT method (Schmittgen and Livak, 2008). All the experiments were repeated thrice with the inclusion of no template control (NTC) for the detection of DNA contamination.

The full-length gene was amplified and cloned in-frame with C-terminal of green fluorescent protein (GFP) in vector pCAMBIA1302 and recombinant vector carrying GFP: StTNLC7G2 construct was introduced into Agrobacterium GV3101 and incubated with onion peel epidermis (Sun et al., 2007). After 2 days, the epidermal layer was thoroughly washed with distilled water and visualized under a fluorescence microscope (Evos, Thermo Fisher Scientific) at 40×.

Conidial suspension of A. solani was prepared using 0.01% Tween-80, and 50 μl was pre-inoculated (painted) on the proposed infiltration site and dried afterward for 30 min. The gene was cloned in pCAMBIA1302 using NcoI and BstEII (Supplementary Figure S2B) and confirmed by sequencing (accession no. MZ597472). The positive Agrobacterium transformants were used for culturing in 200 ml of yeast extract peptone broth (YEP) media supplemented with rifampicin (25 μg/ml) and kanamycin (50 μg/ml) at 28°C to an OD₆₀₀ of ~1.0. The culture was further centrifuged at 4,000×g for 10 min, and the pellet was resuspended in 10 mM of MES [2-(N-morpholino)ethanesulfonic acid] pH 5.6 and 10 mM of MgCl₂ with 100 µM of acetosyringone (Bhaskar et al., 2008) and incubated for 4–8 h at room temperature (RT) in the dark, and 200 µl of the inoculum was used for infiltration of 30-day-old potato leaves of KJ, KC, and KP. The Agrobacterium transformed with a blank vector was used as vector control. The inoculated plants were kept in the dark for 12 h, then shifted in a plant growth chamber with conditions described earlier, and sampled at 3 dpi. For DAB staining, 50 mg/100 ml of working stock was prepared in 0.1 M of sodium phosphate buffer at pH 3.0 and was maintained by using concentrated HCl. Earlier collected samples were now dipped into the staining solution and incubated at RT for 6–8 h at 100 rpm. Then leaves were decolorized using the bleaching solutions (ethanol:acetic acid:glycerol, 3:1:1) for 30 min at 90°C, then replaced with fresh bleaching solutions, and photographed for any brown color appearance denoting ROS (Liu et al., 2014). Gene expression after agroinfiltration was confirmed in N. benthamiana leaves collected at 5 dpi. Samples were collected by an incision measuring 1 × 1 cm² using a razor blade surrounding the infiltration zones (2–3 cm away from the infiltration zones). RT-PCR was performed as previously described (Bhaskar et al., 2008).

The application of Hmmsearch method using Pf00931-pfam domain (NBS) resulted in 644 NBS coding transcripts. The application of TIR Pfam id-1582 within these transcripts narrowed down the result to 110 TIR-NBS-containing transcripts. After assessment of LRR motifs, in total, 55 transcripts were identified containing all the three domains, namely, TIR, NBS, and LRR. As there is always a possibility to skip the detection of the targeted domain by HMM profile due to partial or truncated sequences, a second strategy of CD search using the entire proteome set from the PGSC database was submitted on the NCBI CDD search engine. The strategy provided a total of 60 full-length TNL peptides with a high confidence level for TIR, NBS, and LRR domain encoded by 44 TNL genes (Supplementary Table S2). When these 44 genes were used to identify all possible coded peptides, a total of 36 additional transcripts were identified including 29 truncated sequences. Twenty-seven genes were found to code for more than 1 transcript probably by alternative splicing events (Supplementary Table S3). In MEME analysis, conserved and signature motifs of the TIR domain and NBS domain were identified to be preserved (Supplementary Table S4).

Nomenclature has been assigned to these TNL genes for ease of understanding with an effort to bring uniformity of methodology in such study after retrieving relevant information in a smooth and easy manner. Here, the nomenclature of individual genes and their respective proteins is based on their physical location on the chromosome. For instance, PGSC0003DMG400022699 localized on chromosome 1, at the first position, so the nomenclature of this gene and encoded protein is StTNLC1G1 and StTNLC1G1P1, respectively. If a gene encodes more than one transcript, the nomenclature used here is based on their peptide length in descending order. For example, PGSC0003DMG400024055 encodes 2 peptides: PGSC0003DMP400041617 and PGSC0003DMP400041616. The nomenclature of these peptides will be StTNLC1G2P1 and StTNLC1G2P2, respectively (Supplementary Table S2).

Phylogenetic grouping of the genes from the same chromosomes and the separate chromosomes was observed. Exon–intron structure tends to remain the same among genes in the same clade, indicating significant gene structure conservation over evolution, according to a phylogenetic tree–gene structure correlation. Clades were also formed on the basis of the presence and the absence of a particular motif in proteins and homology between them. Two major clades were formed, i.e., Clade I and Clade II, in which TNLs were demarcated on the basis of the domain and motif compositions (Figure 1A).

The physical location of the TNL genes was resolved on the basis of PGSC_DM_v3.4_gene.gff file as retrieved from the PGSC database. Forty-two out of forty-four TNL genes were successfully mapped on the twelve potato chromosomes using Ritchie Lab phenogram, while only 2 sequences remained unanchored. The chromosomal location of these genes has shown the uneven distribution of TNLs on different chromosomes. Two gene clusters were found, one each on chromosomes 1 and 11 (Figure 1B), both having 6 TNL genes. The number of TNLs mapped on chromosomes 1, 2, 4, 5, 6, 7, 8, 9, 11, and 12 was 9, 1, 3, 4, 6, 4, 1, 3, 10, and 1, respectively. There were no TNL genes found on chromosomes 3 and 10.

Across the potato genome, there were 7 pairs of TNL-TNL segmental duplication events contributed by a total of 9 TNL genes. StTNL36348 was found to be involved in three TNL-TNL events with pairing to StTNL17113, StTNL18619, and StTNL23175 in the genome. Two genes, StTNL37911 and StTNL4310, are probably segmentally duplicated at least six times, including non-TNL duplications. The search for duplication events of TNL vs. other genes identified 35 duplication events/pair. Collinear pair was found on all chromosomes except chromosomes 3, 8, and 12 (Figure 1C). Only one gene, StTNL45901, physically located on chr9 was observed to be tandemly duplicated (Supplementary Table S5).

Before proceeding for expression assessment of transcripts, the virulence of A. solani was confirmed by visualization of symptoms using pathogen spray to plants in pots (Supplementary Figure S4A) and detached leaf assay (Supplementary Figure S4B) at 5 dpi. Both the experiments confirmed the susceptibility of KC to virulent A. solani, while KJ and KP showed tolerance. The results were even further supplemented with fungal biomass assessment (Supplementary Figure S4C). Fourteen genes were chosen randomly for expression analysis, but also keeping in mind that at least one gene from each chromosome may be represented. The differential expression pattern obtained for different transcripts can be summed up in three major patterns for TNL transcripts. The first groups of transcripts have their expression level constantly reaching more than 4.0-fold in expression across all the three varieties (Supplementary Tables S6–S8). Such expression pattern was obtained for seven transcripts, namely, StTNLC1G3T1, StTNLC6G4T1, StTNLC7G1T1, StTNLC9G1T1, StTNLC11G9T1, StTNLC12G1T1, and StTNLCnG2T1. All of them showed more than 10-fold expression in at least one disease-tolerant variety with the exception being StTNLCnG2T1 and StTNLC7G1T1 (Figure 2A). Strong and constant upregulation under pathogen treatment suggests that their expression modulates under pathogenic signals like pathogen-associated molecular pattern (PAMP). A few transcripts showed a difference in expression pattern between tolerant and susceptible varieties. Transcripts with such expression behavior show a constant decrease in expression in KC after the initial spike at 1 or 3 dpi. These transcripts were StTNLC6G2T1, StTNLC7G2T1, StTNLC9G3T1, StTNLC11G7T1, and StTNLC11G8T1. The expression of all these transcripts was decreased to <2.5-fold in KC at 7 dpi while maintaining >6.0-fold expression in KP and KJ.

Meanwhile, two transcripts showed expression of >15.00-fold under A. solani infection. These transcripts were StTNLC6G2T1 and StTNLC11G9T1. The difference in expression value of these two genes at 7 dpi was almost 4 times in the comparison of KJ with KC. StTNLC6G2T1 and StTNLC11G9T1 must have prominent roles in plant defense under A. solani attack at a late stage. For the majority of the genes tested, the expression enhancement pattern was almost similar for KJ and KP with only a significant difference being obtained in StTNLC12G1T1. Furthermore, among all the candidates tested, the expression pattern of only one transcript (StTNLC7G1T1) was found to be constant across the varieties and time points.

One gene, StTNLC7G2, showing homology with N gene of tobacco was selected for further characterization, as its transcript induction was greater than 10-fold at 5 and 7 dpi in KJ and KP, respectively, while expression in KC barely crosses 2-fold with respect to control, though immediately after pathogen inoculation at 1 dpi, the expression was similar in all three varieties. To check the localization of StTNLC7G2, the transcript was amplified and cloned in pCAMBIA1302 in the same frame as of GFP, and positive clones were used to inoculate onion peel. Amplification and cloning of full-length StTNLC7G2 achieved 1,296 bp of amplicon encoding a protein of 431 amino acids (Supplementary Figure S2C). The localization of construct was compared with vector control (GFP only) that clearly showed nucleus localization, while StTNLC7G2 protein showed putative membrane localization (Figure 2B). StTNLC7G2 protein is probably involved in the reception of the pathogenic signal and their further transmission downstream. The high expression of the transcript at 7 dpi in a tolerant variety further indicates the same.

After inoculation of the fungal pathogen, A. solani suspension in the leaves, agroinfiltration was performed with both positive gene construct and vector control at the same site and left for visual observation followed by sampling. DAB staining of 3 dpi agroinfiltrated leaves showed dark patches in all the three varieties at 3 dpi (Figure 3A), indicating the prominent generation of ROS in the presence of StTNLC7G2 mimicking HR (Figure 3B; Supplementary Figure S4D). KJ showed light-dark patches in mock-treated (empty vector) leaves too, suggesting a stronger defense mechanism available to its genotype. An increase in ROS induction in the presence of StTNLC7G2 was more prominent in KP and KC. As ROS induction is the primary feature of HR during plant defense, the StTNLC7G2 is probably co-related with innate immunity against fungal pathogen A. solani. The experiment was further conducted in N. benthamiana leaves with a similar result of ROS induction at 3 dpi (Figure 3C). Semi-quantitative RT-PCR confirmed the expression of StTNLC7G2 in ROS-induced leaves (Figure 3D).