Introduction

Crops are exposed to many diseases, which cause substantial economic losses worldwide. Breeding for disease-resistant varieties offers the most economic and environmentally friendly means of disease control. To date, breeders have focused mainly on the introgression of resistance genes in elite genotypes. These genes, however, often confer race-specific resistance and are not durable because they are frequently overcome by a new, virulent race of a pathogen; in contrast, the silencing or mutating of susceptibility genes, which pathogens require to establish a compatible interaction with the host, are expected to lead to durable resistance1.

The DEFENSE NO DEATH 1 (DND1) gene encodes a cyclic nucleotide-gated ion channel protein2. It was discovered in Arabidopsis thaliana that mutants in the DND1 locus are defective in hypersensitive response but exhibit enhanced resistance against a broad spectrum of virulent fungal, bacterial, and viral pathogens in correlation with the elevated levels of salicylic acid (SA) compounds and mRNAs for pathogenesis-related (PR) genes. As a pleiotropic effect of the above changes these mutants also exhibit a dwarfed rosette phenotype3. Interestingly, mutations that affect SA accumulation or signalling abolish the enhanced resistance of dnd1 mutants against Pseudomonas syringae and Hyaloperonospora parasitica but not Botrytis cinerea; this suggests that the broad-spectrum resistance of dnd1 mutants is related to the activation or sensitization of multiple defense pathways4. DND1 conducts Ca2+ into cells and links the Ca2+ flow to the downstream production of nitrogen oxide (NO), which is an essential signalling molecule invoking plant innate immune response to pathogens5. NO is a central component of the plant senescence signalling cascades. Investigation of dnd1 mutants revealed an interrelationship between Ca2+ and NO generation in leaf cells during senescence. Endogenous NO content in dnd1 leaves is lower than in leaves of wild-type, and these plants show a series of early senescence phenotypes6,7. Besides early senescence, dnd1 mutant Arabidopsis plants flower significantly later, indicating the dependence of flowering time on Ca2+ signalling8. This phenotype, however, is independent of SA or SA signalling9.

The essential role of DND1 in resistance to phytopathogens has been established in non-model and crop plants, too. Homologs of the Arabidopsis DND1 are expressed early during infection by the rust fungus Hemileia vastatrix in coffee10. In the potato and tomato, down-regulation of DND1 leads to resistance to late blight (Phytophthora infestans) and to powdery mildew species, Oidium neolycopersici and Golovinomyces orontii, and impedes the conidial germination, attachment, and hyphal growth of Botrytis cinerea11,12. The tetra-allelic DND1 gene-edited potato lines confer increased late blight resistance. These lines, however, like the DND1 silenced lines11, show reduced growth and leaf necrosis13. Thus, it was concluded that due to the pleiotropic phenotypes observed, DND1 is not a good candidate for application in agriculture13. Nevertheless, DND1 may activate multiple defense pathways in the potato, one or more of which may be separable from the one with the pleiotropic effect, as was demonstrated in A. thaliana4.

To get a general overview on the molecular basis of the pleiotropic effect of DND1 silencing this study aimed to obtain knowledge at three levels and detect transcriptomic, hormonal, and secondary metabolite changes in potato leaves. Here, we report the elevation of SA concentration, the upregulation of 1138 genes, and the downregulation of 728 genes in three DND1 silenced potato lines generated earlier11. The effects of transcriptional and metabolite changes on the growth and fungal resistance of DND1 silenced lines are discussed.

Results

Re-testing and targeted hormone and metabolite analysis of the DND1 silenced potato lines

The DND1 silenced lines DND1-5, DND1-8, and DND1-17 were transferred from Wageningen University, The Netherlands, to our laboratory. To test whether, after propagation in vitro, the transferred lines retained their reduced level of DND1 expression11, RNA was isolated from the leaves of in vitro plants, and the level of DND1 mRNA in comparison to that of the non-transformed control ‘Désirée’ (DES) was tested with RT-qPCR. A 50–60% reduction in DND1 expression was detected in the DND1 silenced lines (Fig. 1).

Fig. 1
figure 1

Level of DND1 expression in DND1 silenced lines DND1-5, DND1-8 and DND1-17 compared to the non-transformed control ‘Désirée’ (DES). RNA was isolated from the middle leaves of in vitro plants; three leaves/plants were harvested from three plants/lines. The RT-qPCR analysis was carried out using the EF1α as a reference gene. The Y-axis shows the average relative 2−ΔΔCT values compared to the average 2−ΔΔCT value of DES set as 1.0. The averages were calculated from three technical replicates. The standard deviations are indicated by the error bars. Significant differences between the DND1 silenced lines and DES were determined by Student’s t-test and labelled by ** (p < 0.01).

Full size image

Mutation in DND1 increases the SA content in A. thaliana3. To test whether this is also the case in the potato, the SA concentration of leaves of in vitro-grown DND1 silenced plants was measured and compared to that of the control DES. A tremendous increase in SA concentration was detected in each DND1 line; while the SA content of DES was 127 ± 18 ng g−1 fresh weight (FW), it was elevated to 2723 ± 1093, 1178 ± 782, and 1333 ± 416 ng g−1 FW in DND1-5, DND1-8, and DND1-17, respectively (Fig. 2). In addition to SA, the phytohormones jasmonic acid (JA) and abscisic acid (ABA) regulate plant defenses through synergistic and antagonistic actions14. An elevated level of JA and its leucine conjugate was detected in the leaves of DND1 silenced plants (Fig. 2). In contrast, no alteration in ABA level was observed in DND1 lines (Supplementary Fig. 1).

Fig. 2
figure 2

Concentration differences of selected hormones and secondary metabolites between the non-transformed control ‘Désirée’ (DES) and the DND1 silenced lines DND1-5, DND1-8 and DND1-17. The data were obtained from four biological replicates from the leaves of each line. Each biological replicate contained the leaves of five in vitro plants. The standard deviations are indicated by the error bars. Significant differences between the DND1 silenced lines and DES were determined by Student’s t-test and labelled by * (p < 0.05) and ** (p < 0.01).

Full size image

Since secondary metabolites, especially those synthesized in the phenylpropanoid pathway, have an important role in plant defense against pathogens15, a targeted metabolite analysis of potato leaves extending to nine compounds was performed. An increased level of chlorogenic acid and its derivatives, as well as that of para-hydroxybenzoic acid, was found in DND1 silenced lines, whereas the concentrations of taxifolin and rutin were decreased (Fig. 2). The silencing did not influence the levels of dihydrokaempferol, phaseic acid, and dihydrophaseic acid (Supplementary Fig. 1).

Transcriptome analysis of the DND1 silenced potato lines

Transcriptome analysis was performed to understand the influence of DND1 silencing on gene expression. RNA was isolated from the leaves of in vitro-grown plants in two biological replicates. The RNA-seq produced good-quality data (Supplementary Table 1), resulting in close to 90% of total and unique mapped reads. The correlation coefficients between the two biological replicates varied from 0.97 to 0.99 overall (Supplementary Table 2). Expression of 18,752 genes was detected in all samples (Fig. 3a). The number of differentially expressed genes (DEGs) compared to DES (Supplementary Tables 3–5) was very similar in DND1-8 (1629 up, 1608 down) and DND1-17 (1744 up, 1347 down), while it was a bit higher in DND1-5 (2146 up, 1847 down). Data are visualized with bar graphs in Fig. 3b and with volcano plots in Supplementary Fig. 2. The number of common genes upregulated in each DND1 silenced line was 1138 (Fig. 3b), whereas the number of downregulated genes was 728 (Fig. 3c).

Fig. 3
figure 3

The number of unigenes (A) and differentially expressed genes (DEGs) (B,C) in the leaves of the non-transformed control ‘Désirée’ (DES) and the DND1 silenced lines DND1-5, DND1-8 and DND1-17. (B) The number of up- and downregulated DEGs in the DND1 silenced lines as compared to DES. (C) Venn diagram of up- and downregulated DEGs. The overlaps represent the genes differentially expressed in more than one line. The number of DEGs detected in each DND1 silenced line is highlighted in yellow.

Full size image

Gene ontology (GO) analysis revealed 57 significantly altered terms: 31 biological processes, 23 molecular functions, and three cellular components. The highest number of genes belonged to the “catalytic activity,” “membrane,” “ion binding,” “response to stimulus,” and “small molecule binding” categories. Interestingly, only five out of 57 categories showed downregulation, namely, “response to chemical,” “response to organic substance,” “cellular response to organic substance,” “cellular response to endogenous stimulus,” and “trehalose metabolism in response to stress” (Fig. 4).

Fig. 4
figure 4

Bubble plot diagram of the significantly (Benjamin and Hochberg FDR correction: p ≤ 0.0001 and REVIGO: simrel 0.7) up- and downregulated GO terms common in the DND1 silenced lines DND1-5, DND1-8, and DND1-17 compared to the non-transformed control ‘Désirée’. BP, biological process; CC, cellular component; MF, molecular function; FDR, false discovery rate.

Full size image

Kyoto Encyclopedia of Genes and Genomes (KEGG)16 analysis was used to further refine the DEG categories by sorting the genes into different metabolic pathways (Fig. 5, Supplementary Tables 6 and 7). It was found that the greatest changes occurred in carbohydrate, lipid, and amino acid metabolism followed by the biosynthesis of other secondary metabolites. For example, β-FRUCTOFURANOSIDASE CELL WALL ISOZYME, SUCROSE SYNTHASE, different types of CHITINASES, PECTATE LYASE, GLUTAMATE DECARBOXYLASE, FATTY ACID DESATURASE, LIPOXYGENASE, GLUTAMATE DECARBOXYLASE, ASPARTATE AMINOTRANSFERASE, and a large number of GLUTATHIONE TRANSFERASES were highly activated. While the number of upregulated and downregulated genes was similar in carbohydrate, lipid, and amino acid metabolism, the extent of changes was, in general, less in the repressed genes than in the activated genes. In the secondary metabolite category, upregulation was the dominant tendency; 41 genes involved in secondary metabolite synthesis, including ELICITOR-INDUCIBLE CYTOCHROME P450, GLYCOSYLTRANSFERASES, PHENYLALANINE LYASES, and LEUCOANTHOCYANIDIN DIOXYGENASE, were upregulated, and only ten genes were downregulated. This tendency was similar to the genes encoding proteins functioning in environmental adaptation (45 up and 11 down). The level of expression of genes encoding CYSTEINE PROTEASE, PATHOGENESIS RELATED PROTEINS, HEAT SHOCK PROTEINS, and CALCIUM-BINDING EF HAND FAMILY PROTEINS was significantly increased. Furthermore, substantial changes were detected in the expression level of genes involved in signal transduction related to environmental information processing; 42 genes were upregulated, and 35 genes were downregulated. Several genes related to auxin response (e.g., AUXIN-RESPONSIVE PROTEIN, AUXIN-RESPONSE FACTOR) and ethylene sensitivity (e.g., ETHYLENE-INSENSITIVE3-LIKE 1, 2, and 3) were repressed.

Fig. 5
figure 5

KEGG pathways16 significantly enriched in each DND1 silenced line. The number of up- and downregulated genes belonging to the identified pathways is indicated. Colours from green to red correlate with the number of genes between 1 and 60.

Full size image

Protein–protein interaction network prediction

Interactions of potato proteins can be predicted based on the STRING database. To do so, the identified potato protein sequences were downloaded from the STRING database, and SpudDB identifiers and sequences were assigned to them. From the common DEG list of the DND1 lines containing 1866 genes in total, 325 corresponding proteins (nodes) with 468 connections (edges) were identified; of them, 230 proteins with 381 connections are found in the largest connected cluster (Fig. 6 and Supplementary Table 8). Within this cluster, several centrals can be found with the most interacting proteins and their primary neighbours. These central proteins possess very diverse functions, including histones, kinases, ubiquitin family proteins, heat shock proteins, and brassinosteroid signalling pathway proteins.

Fig. 6
figure 6

Predicted interactions between potato proteins encoded by differentially expressed genes in DND1 silenced lines. The colour of the circles is associated with gene expression (log2 fold change), while their diameter is correlated with the number of connections. The interacting network contains 325 nodes and 468 edges. Nodes with at least ten connections are numbered, namely: (1) Soltu.DM.01G021410 [histone H3.2-like], (2) Soltu.DM.03G012070 [polyubiquitin-like], (3) Soltu.DM.03G025860 [protein NBR1 homolog], (4) Soltu.DM.03G012080 [polyubiquitin-like], 5) Soltu.DM.10G020640 [heat shock cognate 70 kDa protein 2], (6) Soltu.DM.09G002330 [heat shock cognate 70 kDa protein 2-like], 7) Soltu.DM.03G023440 [heat shock protein 83], (8) Soltu.DM.05G025830 [probable small nuclear ribonucleoprotein G], (9) Soltu.DM.12G005470 [BRASSINAZOLE-RESISTANT 1-like], (10) Soltu.DM.12G022120 [histone H2A.6-like], (11) Soltu.DM.04G034930 [BRASSINAZOLE-RESISTANT 1-like], (12) Soltu.DM.01G039150 [histone H2A.1], (13) Soltu.DM.12G007030 [cryptochrome-1-like]. The source code (graphml) of the graphs is deposited into GitHub Gist and accessible online via yEd Live (https://bit.ly/3tLMqbH).

Full size image

Transcription factors differentially expressed in the DND1 silenced potato lines

Since transcription factors (TFs) are key regulators of stress responses, differentially expressed TFs were analyzed first. Of the TFs that were differentially expressed in each DND1 silenced line, we identified 60 upregulated and 63 downregulated genes (Fig. 7). The WRKY family genes (WRKY6, 13, 16, 24, 30, 40, 41, 48, 51, 54, 70, 75) dominated the group of upregulated genes with the highest increase in expression of WRKY40 (average log2 fold change: 9.9). Several members of the NAC TF family (NAC008, 31, 35, 42, 73, 82, 86, 90) and bHLH TF genes (bHLH13, 29, 35, 90) were also present in the upregulated group. The group of downregulated genes was dominated by the genes encoding bHLH TFs (bHLH8, 50, 51, 58, 63, 93, 104, 121, 130, 137, 151), out of which bHLH8, 151 and 63 had the strongest repression, with average log2 fold changes of − 3.6, − 3.4, and − 2.2, respectively.

Fig. 7
figure 7

Differentially expressed transcription factors (TFs) in the leaves of the DND1 silenced lines DND1-5, DND1-8 and DND1-17 compared to the non-transformed control ‘Désirée’ leaves. The intensity of colours indicates the degree of the difference (log2 fold change).

Full size image

Pathogenesis-related genes differentially expressed in the DND1 silenced potato lines

With the exception of three slightly downregulated pathogenesis-related genes, all of the others were upregulated (Fig. 8). Sixteen categories of upregulated genes could be established. Most of them were related to fungal resistance/sensitivity against, e.g., leaf rust (LRK10Ls, Lr10s), powdery mildew (Mlos), downy mildew (DMR6s), and late blight (R1Bs). Several pathogenesis-related protein genes (PRPs), as well as other disease resistance protein genes (RPs) or probable disease resistance protein genes and hypersensitivity-related protein genes (HSRs), were also upregulated. Additionally, several alleles of the gene encoding the TMV resistance protein N were expressed at a higher level in the leaves of DND1 silenced plants than in DES leaves.

Fig. 8
figure 8figure 8

Differentially expressed pathogenesis-related genes in the leaves of the DND1 silenced lines DND1-5, DND1-8 and DND1-17 compared to the non-transformed control ‘Désirée’ leaves. The intensity of colours indicates the degree of the difference (log2 fold change).

Full size image

Validation of RNA-seq results

Transcriptome data were validated via RT‒qPCR analysis of three upregulated and three downregulated genes. The upregulated genes tested were the TFs WRKY40 and NAC90 and the powdery mildew resistance gene MLO-LIKE PROTEIN 6 (MLO6). The downregulated genes tested were 14 kD PROLINE-RICH PROTEIN (PRO), NADH NITRATE REDUCTASE 3 (NR3), and the TF bHLH8. The expression trends of the genes using RNA-seq and RT‒qPCR were similar in each DND1 line. Further, the correlation coefficients of R2 = 0.6878 in DND1-5, R2 = 9466 in DND1-8, and R2 = 0.9026 in DND1-17 confirmed the reliability of the transcriptome data (Fig. 9).

Fig. 9
figure 9

Validation of RNA-seq results by correlation with RT‒qPCR of six genes. The average log2 fold change in expression levels detected by RNA-seq of the two biological replicates is compared to the average log2 fold change in expression detected by RT‒qPCR of three technical replicates of samples originating from independently grown in vitro plants. The selected genes were as follows: PRO (Soltu.DM.08G025180; green), NR3 (Soltu.DM.11G010380; pink), bHLH8 (Soltu.DM.01G047820; blue), NAC90 (Soltu.DM.11G022040; yellow), MLO6 (Soltu.DM.03G013460; red), WRKY40 (Soltu.DM.08G015910; purple).

Full size image

Effect of exogenous SA on the expression of selected genes

SA is a hormone molecule that can be found at a wide range of endogenous levels in plants and induces responses to various stresses17. In A. thaliana, SA induces the expression of WRKY4018, PR119, MLO620, and DMR621; these genes were highly expressed in the DND1 silenced lines with high SA content. Although no data on the SA-triggered NAC90 transcription activation was found in the literature, this TF was also highly expressed in all three DND1 lines (Fig. 7). To test whether SA can also activate the expression of these genes in the potato, the foliage of greenhouse-grown DES plants was sprayed with SA solution, and leaf samples were collected 24 h after treatment. RT‒qPCR analysis showed that all five selected genes (WRKY40, NAC90, PR1, MLO6, and DMR6) were strongly SA-inducible (Fig. 10).

Fig. 10
figure 10

Effect of exogenous salicylic acid (SA) treatment on the expression of five genes in the leaves of greenhouse-grown ‘Désirée’ plants. The relative expression level of the genes compared to the expression level of ACTIN was determined by RT‒qPCR in three technical replicates. Three plants were treated with 0.6 mM SA as described in the Materials and methods. The standard deviations are indicated by the error bars. The statistical significance of the measurements was determined by Student’s t-test and labelled by ** (p < 0.01). WRKY40 (Soltu.DM.08G015910) and NAC90 (Soltu.DM.11G022040) are transcription factors whereas PR1 (PATHOGENESIS-RELATED PROTEIN 1; Soltu.DM.01G045900), MLO6 (MLO-LIKE PROTEIN 6; Soltu.DM.03G013460) and DMR6 (DOWNY MILDEW RESISTANCE 6; Soltu.DM.06G028410) are related to pathogenesis.

Full size image

Discussion

Previously, it was shown that the silencing of DND1 leads to resistance to late blight, powdery mildew, and gray mold diseases11,12. Using the same lines tested earlier for fungal resistance, our experiment detected an approximately 10- to 20-fold increase in the SA concentration in leaves; this is in good agreement with the 15-fold increase detected in dnd1 mutant Arabidopsis plants3. It is generally considered that SA participates in activating of plant defense mechanisms and acts with other signalling molecules, such as JA22. Besides the elevation of SA content, a 1.5- to 2.0-fold increase in JA and JA-Leu conjugate levels was also observed in the DND1 silenced lines. JA, along with its conjugated forms, tunes plant defense mechanisms by regulating the expression of JA-associated genes imparting the resistance phenotype23. It was demonstrated that JA-Leu can serve as a ligand to promote the interaction between CORONATINE INSENSITIVE1 (COI1) and JASMONATE-ZIM (Zinc-finger Inflorescence Meristem) domain (JAZ) repressor proteins inducing JAZ’s degradation24.

SA- and JA-driven elicitation of secondary metabolites, including phenolics and flavonoids, was detected in various plant species25. In line with this general observation, elevations of concentrations of chlorogenic acid (ester of caffeic acid and quinic acid) and its derivatives were detected in the leaves of DND1 silenced lines compared to the control DES. Thus, the demonstrated antimicrobial effect of chlorogenic acid and related compounds26 may contribute to the fungal resistance of DND1 lines. In contrast, the concentration of the flavonoid taxifolin (also known as dihydroquercetin) and the flavonoid glycoside rutin were lower in DND1 lines than in DES; this indicates that these branches of the phenylpropanoid pathway are repressed due to the silencing of DND1.

SA is synthesized via two pathways in plants: the isochorismate synthase (ICS) and phenylalanine ammonia-lyase (PAL) pathways27. Isotope feeding experiments suggest that SA is synthesized by PAL from phenylalanine via trans-cinnamic acid and benzoic acid in the potato28; however, the existence of the ICS pathway cannot be excluded. We found that not only the level of SA but also the amount of para-hydroxybenzoic acid (PBHA) was significantly increased in DND1 leaves. Because benzoic acid is a precursor of SA in the PAL pathway, the increased PBHA level detected in DND1 lines may support the previous finding that the bulk of SA is synthesized via PAL in the potato. Based on cDNA heterogeneity, it was estimated that at least about ten, but probably more, PAL genes are active in the potato29. We found eleven, nine, and eight PAL genes in DND1-5, DND1-8, and DND-17, respectively, with higher expression levels than in DES, out of which eight PAL alleles were common (Supplementary Table 9); this indicates a negative correlation between DND1 function and PAL expression.

Based on Arabidopsis studies SA level is regulated by both positive and negative feedback. Briefly, ICS1 expression in Arabidopsis is controlled mainly by SYSTEMIC ACQUIRED RESISTANCE DEFICIENT 1 (SARD1) and CALMODULIN BINDING PROTEIN 60 g (CBP60g) TFs. The TFs WRKY8, 28, 46, 48, and 75 and TEOSINTE BRANCHED1/CYCLOIDEA/PCF (TCP)-binding site TFs TCP8 and TCP9 promote ICS1 expression, while ANAC019, 055, and 072, BENZOIC ACID/SA CARBOXYL METHYLTRANSFERASE 1 (BSMT1), ETHYLENE INSENSITIVE 3 (EIN3) and EIN3-LIKE 1 (EIL1), and WRKY18, 40, 54, and 70 repress it30. Out of these TFs, WRKY48, WRKY75, and TCP8 were upregulated in DND1 silenced potato lines with average fold changes of 39, 5.7, and 2.1 (Fig. 7), respectively, whereas only EIL1 of the negative regulators appeared among the downregulated genes. Moreover, the level of ICS was not changed in DND1 leaves compared to DES leaves, as it was not present among the DEGs, indicating that the ICS pathway of SA synthesis is much less significant in the potato than in Arabidopsis. Expression of WRKY48 and WRKY75 was connected to the JA signalling in Catharanthus roseus and the tomato, respectively31,32, and this may also be the case in the potato.

In sum, 1866 common DEGs were detected in the DND1 lines, and more than 60% (1138) of them were upregulated. The dominance of upregulation was particularly striking in the case of significant GO terms when it reached 91% (52/57). KEGG analysis pointed out that 80% (41/51) of the DEGs related to the biosynthesis of secondary metabolites were activated. Phenylpropanoids are derived from phenylalanine, and the first enzyme in the metabolon is PAL33. Eight alleles of PAL are induced at 1.9- to 6.6-fold (Supplementary Table 9) in DND1 lines, leading indirectly to the increase in the amount of chlorogenic acid and its derivatives (Fig. 2). Although we did not test the concentrations of flavonoids, several genes involved in flavonoid biosynthesis, e.g., LEUCOANTHOCYANIDIN DIOXYGENASE (72-fold), CAFFEOYL-CoA O-METHYLTRANSFERASE (22-fold), and FLAVONOID 3′,5′-HYDROXYLASE (4-fold), showed an elevated level of expression; this suggests an increase in concentration of a high range of flavonoids.

The most prominent enzymatic elements of the reactive oxygen species (ROS)-activated antioxidant plant defense machinery are PEROXIDASE, CATALASE, and SUPEROXIDE DISMUTASE34. Interestingly, while the mRNA levels of PEROXIDASES increased by 5- to 26-fold, there was only a 1.4- to 1.8-fold increase in the expression of SUPEROXIDE DISMUTASE; a 1.4- to 1.5-fold reduction was detected in the transcript level of a CATALASE. Thus, we concluded that the silencing of DND1 might not activate the antioxidant defense mechanism and the elevated level of PEROXIDASES might rather contribute to the increased phenypropanoid synthesis than to ROS elimination.

As in the case of the secondary metabolite metabolism, 80% (45/56) of the DEGs related to environmental adaptation were activated in the DND1 lines. These include numerous alleles of Mlo-like, LRK10L, Lr10, DMR6, and R1B genes involved in plant-fungi interactions. However, members of the MILDEW LOCUS O (MLO) gene family act as susceptibility factors by recognizing pathogens, and the corresponding loss-of-function mutations confer broad-spectrum durable host resistance35. Furthermore, the mutated form of DMR6 of Arabidopsis shows resistance to downy mildew36. Thus, it is unlikely that the elevated expression of Mlo-like or DMR6 would confer the reported late blight, powdery mildew, and gray mold resistance in the DND1 lines. It is much more likely that the upregulation of Lr10s, LRK10Ls, and R1Bs, which are all resistance genes with proven effects against leaf rust in cereals and late blight in potato37,38,39, contribute to the fungal resistance of the DND1 silenced lines.

Infection of potato leaves with the late blight pathogen P. infestans leads to the massive accumulation of PR proteins40, and several alleles of PR genes showed extreme expression in DND1 lines. PR-1 expression was up to more than 800-fold (log2 = 9.8 fold) higher in DND1 lines than in the DES control (Fig. 8). Transgenic tobacco plants expressing the PR-1a gene at high levels have long been known to exhibit increased tolerance to P. parasitica and Peronospora tabacina41. Besides PR-1, PR-2, 4, and 6 are activated in the DND1 lines. PR-2 is a β-1,3-glucanase and, in conjunction with PR-1, provides resistance against biotrophic fungi. PR-4 is an endochitinase that aids in the degeneration of cell walls, while PR-6 is a proteinase inhibitor42. It is known that PR-1, 2, and 6 are regulated by SA43. Thus, we assume that the increased level of SA activated the expression of PR genes in the DND1 silenced lines.

The other well-known class of plant resistance proteins is the nucleotide-binding site and leucine-rich repeat domain proteins (NBS-LRRs), which are immune sensors and recognize the pathogen-derived molecules termed avirulence proteins44. In rice, RGA4 and RGA5 act in interaction to mediate resistance to the fungal pathogen Magnaporthe oryzae. RGA4 is a trigger of cell death that is repressed in the presence of RGA5. Upon recognition of the pathogen effector by binding to RGA5, repression is relieved, and cell death occurs45. In the DND1 lines, four alleles of RGA4 showed a two to eightfold increase in expression, while RGA5 was not among the DEGs. Thus, we assume that the uncontrolled activity of RGA4 could be one of the factors causing the leaf necrosis described in DND1 silenced potato lines11. Another factor might be the overexpression of the HYPERSENSITIVITY-INDUCED RESPONSE PROTEIN 1 (HIR1) in DND1 lines, as this defense gene family is associated with hypersensitive reactions involving cell death and pathogen resistance46,47. Notably, eight alleles of TMV RESISTANCE PROTEIN N were upregulated approximately 2- to 30-fold. This gene encodes a typical NBS-LRR protein that localizes TMV infection to cells adjacent to the site of viral entry and develops a hypersensitive response in the form of local necrotic lesions48. Thus, the constitutive expression of TMV RESISTANCE PROTEIN N might be the third reason for leaf necrosis in DND1 lines.

DND1 is a cyclic nucleotide-gated ion channel protein2. How can the reduced level of this protein cause changes in the expression level of almost 2000 genes? To answer this question, we tested the SA inducibility of five genes (WRKY40, NAC90, PR1, MLO6, DMR6) as representatives and predicted protein–protein interactions. All five genes, including the two TFs, turned out to be SA-activated. In total, twelve WRKY and eight NAC TF family members were induced in the DND1 lines. The role of WRKY TFs in regulating plant disease defense signalling is already well demonstrated in several plant species49. In the potato, 110 NAC TF genes were identified, and seven of them were proven to improve the potato plant’s ability to resist Phytophthora infection50,51. Nevertheless, none of these seven genes were identical to those NAC genes that we identified as upregulated genes in the DND1 lines. The bHLH family members dominated the group of downregulated TFs. These TFs have a pleiotropic regulatory role in plant growth and development but are also involved in stress responses52. Thus, we assume that the downregulation of bHLHs would be one of the reasons for the retarded growth rate of DND1 silenced lines. The predicted protein–protein interactions, however, did not show WRKY, NAC, or bHLH TFs as central proteins in the largest connected cluster, but some WRKYs had interactions with other WRKYs, such as, for example, WRKY70 with WRKY50 and WRKY51 (Supplementary Table 8). The detected central proteins included histones, ubiquitin family proteins, kinases, heat shock, and brassinosteroid pathway-related proteins with very diverse and general functions. Thus, further studies would be necessary to elucidate the mechanism of how DND1 influences the expression of approximately 2000 genes and the growth of plants.

Conclusion

Transcriptome, hormonal, and secondary metabolite analysis of the leaves of three in vitro-grown DND1 silenced lines revealed a 10- to 20-fold increase in SA concentration that might be the major cause of transcriptional changes in almost 2000 genes. The upregulated expression of eight PHENYLALANINE AMMONIA-LYASE (PAL) genes and the unaltered level of ISOCHORISMATE SYNTHASE (ICS) mRNA supported the previous finding that, unlike in Arabidopsis, the bulk of SA is synthesized via PAL in potato. In correlation with PAL expression, the concentration of chlorogenic acid and its derivatives was increased, but the antioxidant defense mechanism was not activated. Several resistance genes, including Lr10s, LRK10Ls, R1Bs, and PRs, were upregulated, which might be the main factor behind fungal resistance, while the overexpression of the R-GENE ANALOG 4 (RGA4), HYPERSENSITIVITY-INDUCED RESPONSE PROTEIN 1 (HIR1), and TMV RESISTANCE PROTEIN N might explain the existence of local necrotic lesions in the leaves of DND1 silenced lines. Dominated by WRKYs and NACs, 60 upregulated TF genes were identified, whereas in the group of 63 downregulated TF genes, the bHLHs were in excess. Because bHLHs highly influence plant growth and development, we hypothesize that their downregulation is a key factor in turning the DND1 plants from growth to defense.

Materials and methods

Plant materials and growth conditions

The DND1 silenced lines DND1-5, DND1-8, and DND1-17 were previously generated from Solanum tuberosum cv. ‘Désirée’11 and transferred to our laboratory via authorized transboundary movement. They were propagated in vitro in rooting medium RM (MS53 without vitamins containing 2% (w/v) sucrose and 0.8% (w/v) agar) at 24 °C with a 16-h photoperiod at a light intensity of 75 μmol m−2 s−1 for 4 weeks in 40-mL test tubes closed with paper plugs and transferred into pots containing Tabaksubstrat sterile soil A200 (Stender GmbH, Schermbeck, Germany). The plants in pots were grown under greenhouse conditions at a temperature regime of 18–24 °C and a 16 h/8 h day/night photoperiod. Soil humidity was approximately 80%. Plants were irrigated according to need and treated weekly with acetamiprid-containing pesticide (Mospilan 20 SG).

Salicylic acid treatment

To study the effect of the exogenous salicylic acid (SA) on gene expression, 4-week-old pot-grown ‘Désirée’ plants were treated with 30 mL of 0.6 mM SA54 (S1367, Duchefa Biochemie, Haarlem, The Netherlands) solution using a hand sprayer with a full cone nozzle. For the control treatment, deionized water spraying was applied. Foliar spraying was done at 11:00 am and 3:00 pm. Three plants were subjected to both the SA and control treatments. Twenty-four hours after the first foliar spraying, three leaves from each plant were collected for target gene expression analysis. Leaf samples were frozen in liquid nitrogen and stored at − 80 °C until processing.

Targeted metabolite analysis

Ultra-performance liquid chromatography–tandem mass spectrometry (UPLC‒MS/MS) was used to detect general phenolics and plant hormones in the leaves of 4-week-old in vitro plants as described earlier55. Briefly, frozen leaves were ground in a mortar with a pestle. Samples were spiked with labelled [2H6](+)-cis, trans-abscisic acid as an internal standard and extracted with methanol:water (2:1) with vigorous shaking. Then, samples were filtered through a 0.22-µm PTFE syringe filter and submitted for analysis. Separation was achieved on an Acquity I class UPLC system (Waters, Milford, MA, USA) on a Waters HSS T3 column (1.8 μm, 100 mm × 2.1 mm). Gradient elution was used with 0.1% (v/v) formic acid in both water (A) and acetonitrile (B). Tandem mass spectrometric detection was performed on a Waters Xevo TQ-XS equipped with a UniSpray™ source operated in timed multiple reaction monitoring mode.

Transcriptome analysis

RNA was isolated56 from a pool of 15 leaves harvested from five 4-week-old in vitro plants. Two independent pools were prepared from each DND1 silenced line and the control ‘Désirée.’ RNA samples were transported to the Novogene Company Ltd. (Cambridge, UK) and used for quality control and the generation of sequencing libraries. A 150-bp paired-end sequencing strategy (Illumina NovaSeq 6000 platform) was carried out followed by a quality check of the resulting data. The basic bioinformatic analysis including mapping of quality reads (FPKM, Fragments Per Kilobase Million) to the latest version (v.6.1) of the S. tuberosum group Phureja DM1-3 reference genome57 (http://solanaceae.plantbiology.msu.edu/dm_v6_1_download.shtml), gene expression level analysis (e.g., coexpression Venn diagram; Fig. 3a), and differential gene expression analysis (log2 fold change of DEGs, Supplementary Tables 3–5 and volcano plot, Supplementary Fig. 2) were also performed by the company Novogene.

The functional annotation of the network nodes (DEGs) followed a 3-way approach. First, to determine functional domains from DEGs, their associated amino acid sequences were scanned with the Hidden Markov Model (HMM)-based HMMER v3.3.2 software package58 using Pfam-A v35.0 HMM profiles (ftp://ftp.ebi.ac.uk). Second, for gene ontology (GO) analysis, an up-to-date ontology database (http://geneontology.org/docs/download-ontology; go-basic) and custom-made gene annotation dataset were assembled based on ontology information of the Universal Protein Resource (ftp://ftp.uniprot.org/pub/databases/uniprot/previous_releases/release-2019_04/). The enrichment analysis was performed with the BiNGO v3.0.3 plug-in59 to the Cytoscape v3.7.2 software60. Over-represented GO terms were determined using a hypergeometric statistical test with Benjamin and Hochberg False Discovery Rate (FDR) correction at a p ≤ 0.0001 significance level. Then, REVIGO61 was used with default settings (simrel; 0.7) to reduce the GO term list. Datasets from the GO enrichment analysis were visualized with the ggplot2 R-package (https://github.com/tidyverse/ggplot2). Finally, the Wget software package (https://www.gnu.org/sofware/wget) was used to retrieve the corresponding datasets from the KEGG knowledgebase62, including functional annotations, metabolic pathway classifications, and nucleotide/protein sequences.

Potato-related protein–protein interaction data were retrieved from the STRING database63. Specific identifiers among the various potato genome projects (SpudDB, http://spuddb.uga.edu/index.shtml and the PGSC Ensembl database https://plants.ensembl.org/Solanum_tuberosum/Info/Index) were clarified based on the homology of sequences using the BLASTP tool64. Network topology was designed and visualized with yEd Graph Editor software version 3.23.1 (https://www.yworks.com/products/yed). The source code (graphml) of the graphs was deposited to GitHub Gist and is accessible online via yEd Live (https://bit.ly/3tLMqbH).

Targeted gene expression analysis

The total RNA purified from 150 mg of frozen leaf tissues56 was dissolved in 30 μL of deionized water, and the concentration was measured at OD260 using an ND-1000 Nanodrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

The expression level of DND1 in the silenced lines DND1-5, DND1-8, and DND1-17 compared to the non-transformed control ‘Désirée’ was tested in cDNA samples reverse transcribed from the RNA samples (1000 ng) with a RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific) using random primers. The cDNA samples were diluted threefold, and 0.6 µL was added as a template into 10-µL qPCR reactions. Primers were used in 1 µM final concentration each at a melting temperature of 60 °C together with the Fast SYBR® GreenMaster Mix (Applied Biosystems, Waltham, MA, USA). qPCRs were carried out in triplicates in a Fast 7500 instrument (Applied Biosystems). The housekeeping potato gene ELONGATION FACTOR 1α was used as the reference gene. The transcription level was calculated using the 2−ΔΔCT method65 with efficiency correction according to Pfaffl66.

For validation of RNA-seq results RNA was isolated from the leaves of a new generation of in vitro plants. RT‒qPCR analysis was conducted with 2 μL of total RNA using the GoTaq@ 1-Step RT‒qPCR System kit (Promega Corporation, Madison, WI, USA). Standard cycling conditions of 3-step amplification were started at 37 °C for 15 min and 95 °C for 10 min, then followed by 40 cycles of 95 °C for 10 s, 60 °C for 30 s, and 72 °C for 30 s. To test the effect of SA treatment, the cDNA was synthesized using the Maxima H minus First Strand cDNA Synthesis Kit with dsDNase according to the manufacturer’s (Thermo Fisher Scientific) instructions. RT‒qPCR assays were performed with 2 µL of cDNA using the Xpert Fast SYBR (uni) 2× Master mix (GRiSP Research Solutions, Porto, Portugal) under the following thermal conditions: 95 °C for 3 min followed by 40 cycles of 95 °C for 5 s, 60 °C for 30 s, and 72 °C for 30 s. Both RT‒qPCR experiments were carried out in a Light Cycler-96 Thermal cycler and analyzed using the Light Cycler-96 Software version 1.1 (Roche Diagnostics GmbH, Mannheim, Germany). ACTIN was used as a reference gene67 in both experiments.

Primers (Supplementary Table 10) were designed using the NIH Primer-BLAST tool (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) with default parameters.

Statistical analysis

Calculation of log2 fold changes and correlation parameters was done by using the MS Excel 2023 software. The statistical significance of the measurements was determined by Student’s t-test at the p < 0.01 and p < 0.05 levels.