The NBS region of the NBS-LRR (Nucleotide Binding Site Leucine-Rich Repeats) class of R genes was used to probe the transcriptome of D. sissoo to identify the resistance gene sequence(s) against dieback disease. The NBS region of the NBS-LRR domain consists of conserved domains, such as kinase P-loop, kinase-2, kinase-3A, and hydrophobic GLPL motifs. Based on the NBS domain, degenerate primers were designed and synthesized to amplify resistance gene analogs (RGAs) against the dieback disease of D. sissoo ( Table S1 ). Two primer sets were designed with the support of literature. One set from the 5’ P-loop domain was the forward primers, and the other from the 3’ GLPLA domain was the reverse primers.

A survey across Pakistan was conducted during 2017-2018 to collect healthy germplasm of D. sissoo. Branch cuttings of healthy plants were taken and macropropagated in the greenhouse of the Fungal Molecular Biology Laboratory, Department of Plant Pathology, University of Agriculture Faisalabad, Pakistan. For nursery raising, the cuttings were planted in polythene bags filled with a mixture of soil and sand (3:1 ratio) (Ijaz et al., 2019).

Macropropagated tagged material of D. sissoo was inoculated with Ceartocystis dalbergicans (Haq et al., 2023). Spore resuspension was adjusted to 1 × 10¹¹ spores/ml for inoculations. Most plants died within a month post-inoculation. Several died in 2 months and we re-inoculated the remaining plants (with fewer or no disease symptoms) 3 months post-inoculation. The plants that still showed no symptoms after 6 months were tagged as resistant sources and probed for identifying resistance genes.

Plants showing resistance under the dieback challenge were subjected to RNA isolation. The sampled leaves were immediately kept in liquid nitrogen prior to the RNA isolation procedure. Total RNA was isolated using the GeneJET Plant RNA Purification Mini Kit (Thermo Scientific, USA) following the manufacturer’s protocol. The dried pellet was dissolved in nuclease-free water and immediately used for downstream applications.

The extracted RNA of each sample was quantified on a UV-visible NANODROP (8000 Spectrophotometer, Thermo SCIENTIFIC), and quality was assessed through gel electrophoresis. The RNA samples were subjected to cDNA synthesis. First-strand cDNA was synthesized using the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific, USA), following the manufacturer’s protocol. Degenerate oligonucleotide primed-reverse transcriptase PCR (DOP-rtPCR) was performed to amplify putative disease resistance genes or RGAs (resistance gene analogs) from the transcriptome of D. sissoo under dieback stress. The reaction conditions for the DOP-rtPCR were optimized. Amplicons were resolved on 1.2% agarose gel and visualized under UV on the gel documentation system (Bio-Rad, USA). The required bands were cut and eluted from gel using a gel purification kit (FavorPrep) and sequenced on the Sanger platform (Eurofins Genomics, USA). The generated sequences were characterized in silico. The generated sequences of identified RGAs were translated using the Expert Protein Analysis System (ExPASy) translate tool.

The translated proteins were scanned for structural motifs from the PROSITE collection of motifs. The ScanProsite web tool from ExPASy (https://prosite.expasy.org/scanprosite/) was used for this purpose.

The online web servers CELLO v. 2.5 (Yu et al., 2006) and SignalP 5.0 (Armenteros et al., 2019) were used to predict the subcellular localization of protein or signal peptides. An online ExPASy tool, ProtParam (Gasteiger et al., 2005), was used for computing the physicochemical parameters.

To identify conserved domains in the identified RGAs of D. sissoo, online tools from NCBI, CD-search (Lu et al., 2019), and CDART (Geer et al., 2002) were used. Similarly, protein fingerprints were scanned using the PRINTS database (Attwood et al., 1994).

The secondary structures of deduced amino acid sequences were predicted and annotated through SOPMA (self-optimized prediction from multiple alignments) (Geourjon and Deleage, 1995) and PSIPRED 4.0 (Jones, 1999) online servers. In homology modeling analysis, SWISS-MODEL (Waterhouse et al., 2018) and PHYRE2 (Protein Homology/analogy Recognition Engine, Version 2.0) (Kelley et al., 2015) were employed. The template selection was based on significant score values. Their generated PDB files were used in PyMOL v. 2.4.0 for 3-D protein imaging. Ramachandran plots were generated through PROCHECK (Laskowski et al., 1993) to determine model quality. TM-align (Zhang and Skolnick, 2005) was employed for model superimposition.

Five different motifs were identified in the RGAs of D. sissoo. These five motifs were found with different signatures ( Table 1 ). A protein kinase phosphorylation site was observed in the recovered RGAs.

Subcellular localization prediction analysis was made in CELLO v 2.5. It confirmed their role as a subcellular entity. The physical and chemical properties of the deduced D. sissoo RGA proteins were also computed. These included amino acid composition, molecular formula, molecular weight (Mw), theoretical pI (isoionic point), aliphatic index, instability index (II), and the grand average of hydropathicity (GRAVY) of the deduced amino acids ( Supplementary Table 2 ). The biochemical and molecular functioning of proteins relies on these properties. Amino acid composition confirmed the low abundance of methionine (0.9-1.7%), tryptophan (1.6-1.8%), and aspartate (1.6-3.3%), while there was a higher abundance of leucine (10.5-15.3%), proline (10.5-13.3%), serine (11.9-15.0%), and arginine (8.3-16.7%) in the deduced amino acid sequences. The isoionic or isoelectric point of a protein measures the pH at which its net charge becomes zero or neutral due to an ionizable group, i.e., aspartic acid, arginine, cysteine, glutamic acid, histidine, tyrosine, and lysine. All of these predicted proteins showed the basic range of isoelectric points. While the aliphatic index is based on hydrophobic and non-polar side chain residues, i.e., leucine, isoleucine, valine, and alanine, its high value indicates that the protein is thermally stable. The aliphatic index of the deduced amino acid sequences of the identified RGAs predicted that these proteins are thermally stable ( Supplementary Table 2 ). Stable protein instability index (II) values should be 40 or lower. According to this prediction, the Ds-DbRCaG-02-Rga.a protein was categorized as stable. GRAVY measures the addition of the hydropathy values of the side-chain amino acids, while the sequence length is taken as a divisor. A higher GRAVY score (+ve) for proteins indicates an excellent hydrophobic nature. Among the identified RGAs, DbRCaG-02-Rga.a was predicted to have a hydrophobic nature, scoring a 0.675 value.

The translated proteins of the D. sissoo RGAs had no conserved domains upon searching through CDART and CD programs. The FPScan predicted the protein fingerprints of the translated amino acid sequences of these RGAs. The translated protein sequence of Ds-DbRCaG-02-Rga.a was predicted with an ELONGATNFCT (elongation factor signature) fingerprint. The protein synthesis elongation factor (EF-Tu) binds to the GTP molecule during the translation process and is now thought to be involved in plant stress response. The ELONGATNFCT is a five-element fingerprint that dispenses a signature to GTP-binding elongation factors. These five motifs were derived from the C-terminal site of the initial alignment of 12 elongation factors. The translated protein sequence of Ds-DbRCaG-01-Rga.a had shown homology with motif-2 and motif-4 in the ELONGATNFCT signature.

The translated protein sequence of Ds-DbRCaG-04-Rga.b was predicted with a VDCCGAMMA6 (voltage-dependent Ca²⁺ channel gamma 6) fingerprint. Calcium ion (Ca²⁺) concentration is an essential indicator of signal transduction during plant growth, development, and stress. These consist of four subunits: beta subunit (intracellular); gamma subunit (transmembrane); the cleaved product of the same gene, alpha-2; and delta subunit (Witcher et al., 1993). The VDCCGAMMA6 is a five-element fingerprint that provides signature sequences for the voltage-dependent calcium channel gamma6 subunit. These five motifs were derived from the initial alignment of four sequences. Motifs 1 and 2 reside in the cytoplasmic N-terminal site, motifs 3 and 4 in the first extracellular loop, while motif-5 covers the transmembrane domain-4, including the intracellular segment C-terminal region. The translated protein sequence of Ds-DbRCaG-01-Rga.b had shown homology with motif-1 and motif-3 in the VDCCGAMMA6 signature.

The translated protein sequence of Ds-DbRCaG-06-Rga.c was predicted with three fingerprint sequences. TNFACTORR13C provides signature sequences for necrosis or apoptosis as an activated host defense (Chu, 2013). The TNFACTORR13C is a three-element fingerprint derived from the initial alignment of three sequences based on the conserved region. The extracellular N-terminus region is covered by motif 1, while motif 2 spans at the N terminus and a transmembrane region. However, motif 3 resides in the C-terminus intracellular region. The translated protein sequence of Ds-DbRCaG-06-Rga.c predicted homology to motif-2 and motif-3 signature sequences. The second fingerprint is PROTEASOME provides a signature for proteasome subunits. The accumulation of proteasomes during stress is evidence of enhancing the gene expression of this protein complex. The proteasome first converted damaged protein into a short peptide chain and later to single amino acids used as building blocks of other proteins. The PROTEASOME is a four-element fingerprint derived from the initial alignment of six sequences based on conserved region.

The translated protein sequence of Ds-DbRCaG-06-Rga.c was predicted to be homologous with motif-1 and motif-4 of the PROTEASOME signature sequence. However, the third fingerprint, MTPLANTPEC, provides signature sequences for metallothionein PEC superfamily members. The MTPLANTPEC is a five-element fingerprint derived from the initial alignment of six sequences. Motif-1 has three metal-binding cysteine residues; however, motifs 2 to 5 contain cysteine residues 1, 2, 4, and 3, respectively, for metal binding. The translated protein sequence of Ds-DbRCaG-01-Rga.c had shown a match with motif-2 and motif-5 of the MTPLANTPEC signature.

The SOPMA and PSIPRED programs were used for secondary structure assessments. The α-helix (Hh), random coil (Cc), and extended strand (Ee) percentiles were observed in the translated protein sequence of D. sissoo RGAs. The type of amino acid (polar, non-polar, hydrophobic, and aromatic), annotation grid or sequence plot, and PSIPRED chart were also computed ( Supplementary Figure 1 ). In Ds-DbRCaG-02-Rga.a, a significant share, 63.89%, of the protein secondary structure Ee was recorded, while the percentage of other shared structural features was 2.78% for Hh, 13.89% for Tt, and 19.44% for Cc. The translated protein sequence of Ds-DbRCaG-04-Rga.b predicted 22.8% of Hh, 31.58% of Ee, and 10.53% of Tt, with a significant Cc share of 35.09%. Likewise, the translated protein sequence of Ds-DbRCaG-06-Rga.c predicted 25.98% of Ee and 12.60% of Tt, with a considerable share, 61.4%, of Cc.

Different online servers were used for the structural and functional characterization of the D. sissoo RGAs. The deduced amino acid sequences of the identified RGAs were subjected to SWISS-MODEL and PHYRE2 for template selection.

The Ds-DbRCaG-02-Rga.a showed maximum homology with a significant hit to template 3UAF (the crystal structure of a TTR-52 mutant of Caenorhabditis elegans). Twenty-three amino acid residues of Ds-DbRCaG-02-Rga.a (11-33) were modeled with template 3UAF residues (48-77) with 53.4% confidence and 39% sequence identity. The aligned region of the template confirmed the TTR52 superfamily (Transthyretin-like family) domain.

The Ds-DbRCaG-04-Rga.b showed homology with a significant hit to the template 2W27 (Crystal structure of the Bacillus subtilis YkuI protein an EAL domain, in complex with substrate c-di-GMP and calcium). Forty-six amino acid residues of Ds-DbRCaG-04-Rga.b (57-102) were modeled with template 2W27 residues (306-348) with 29.6% confidence and 21% sequence identity. However, the aligned region of the template confirmed EAL associated YkuI_C domain. This EAL domain is named after the presence of a conserved motif signature, Glutamate-Alanine-Leucine.

The Ds-DbRCaG-06-Rga.c showed homology with a significant hit to the template 3M1U (Crystal structure of a Putative gamma-D-glutamyl-L-diamino acid endopeptidase (DVU_0896) from Desulfovibrio vulgaris Hildenborough at 1.75 A resolution). Twenty-six amino acid residues (30-55) of Ds-DbRCaG-06-Rga.c were modeled with template 3M1U residues (135-160) with 62.6% confidence and 42% sequence identity. However, the template’s aligned region confirmed the SH3 (Src homology 3) domain, which involves regulating kinase activity.

After template selection, 3D imaging of these models was generated using PyMOL v. 2.4.0 ( Figure 2 ). The Ds-DbRCaG-02-Rga.a encoding protein showed structural homology to some parts of the TTR52 domain, consisting of a loop region, as shown in Figure 2A . The crystalline structure of template 2W27 predicted two domains: N-terminal EAL and C-terminal EAL, associated the YkuI_C domain, both connected with long α-helix. The Ds-DbRCaG-04-Rga.b encoding protein showed structural homology to some parts of the YkuI_C domain, typically containing six antiparallel β sheets packed against the helical structure, giving a pocket-like appearance. The crystalline structure of the Ds-DbRCaG-04-Rga.b encoding protein contained three small α-helix and two β-sheets joined via extended loops, as shown in Figure 2B . The crystalline structure of template 3M1U was predicted with four domains: N-terminal c-clip, SH3b1, SH3b2, and C-terminal NPLC-P60. However, the Ds-DbRCaG-06-Rga.c encoding protein showed structural homology to some regions of the SH3b1 domain. The SH3 has a β barrel structure mainly consisting of five or six β sheets. In contrast, the crystalline structure of Ds-DbRCaG-06-Rga.c encoding protein contained a loop region, as shown in Figure 2C .

The quality was assessed by generating Ramachandran plots to confirm the reliability of homology-based generated 3D models. The model validation of Ds-DbRCaG-02-Rga.a predicted 82.4% residues in the most favored region and 11.8% in the additional allowed region with the remainder in the disallowed region. The model validation study of Ds-DbRCaG-04-Rga.b indicated 78.0% residues in the most favored region, 14.6% in the additional allowed region, and 7.3% in the generously allowed region. Similarly, a model validation study of Ds-DbRCaG-06-Rga.c indicated 83.3% residues in the most favored region, 11.1% in the additional allowed region, and 5.6% in the generously allowed region. Moreover, the calculated Z-scores for the predicted models of Ds-DbRCaG-02-Rga.a, Ds-DbRCaG-04-Rga.b, and Ds-DbRCaG-06-Rga.c were -0.16, -2.16, and -1.72, respectively, confirming the models to be satisfactory ( Figure 3 ). The superimposition of the model with the template was also assessed along with the TM score and RMSD (root mean square deviation) values. The TM scores of Ds-DbRCaG-02-Rga.a, Ds-DbRCaG-04-Rga.b, and Ds-DbRCaG-06-Rga.c were 0.13608, 0.10621, and 0.05534, respectively, suggesting random structural similarity of these superimposed models with their template. While the RMSD values of Ds-DbRCaG-02-Rga.a, Ds-DbRCaG-04-Rga.b, and Ds-DbRCaG-06-Rga.c with their respective template were 2.44 Å, 0.73 Å, and 2.22 Å, respectively. The RMSD value scales the similarity among two atomic coordinates in a superimposed position. In the present findings, the measured RMSD values demonstrated that the residues were superimposed well ( Figure 4 ).