Introduction

Host-induced gene silencing (HIGS) takes advantage of the RNA interference (RNAi) pathway, which is conserved within most eukaryotic organisms1,2. This pathway can be triggered by dsRNA molecules recognized and further processed into small interfering RNAs (siRNAs) by Dicer-like proteins (DCLs). These siRNAs with sequence complementarity to target pathogen transcripts are subsequently bound to Argonaute proteins (AGOs), forming the RNA induced silencing complex (RISC) with complementary siRNAs guiding the RISC to target mRNA transcripts for degradation3,4. HIGS uses complementary transgene-expressed hairpin (hp)RNA that is capable of targeting a single gene’s transcripts, thus removing or reducing the risk of off-target effects on the agroecological environment5. Applications of HIGS have demonstrated success against a diverse set of fungal pathogens, including ascomycetes Botrytis cinerea6, Fusarium graminearum7 and Verticillium dahlia8, basidiomycetes Ustilago hordei9 and Melampsora lini10, and oomycetes Phytophthora infestans11 and Hyaloperonospora parasitica12. Successful HIGS-mediated control of a phytopathogen is dependent on choosing a target gene whose transcript abundance can be adequately reduced by the hairpin (hp)RNA-expressing transgene13,14. In prior studies, we explored the efficacy of the putative tRNA synthetase ABHYDROLASE-3 as an RNAi candidate gene through the topical application of long dsRNA in spray-induced gene silencing (SIGS) to reduce pathogenicity of S. sclerotiorium on B. napus and Arabidopsis thaliana15. Significant reductions in S. sclerotiorum lesion progression post application of ABHYDROLASE-3 targeting dsRNA molecules were correlated with reductions in target transcript levels as early as 24 h post infection. Subsequently, Wytinck et al.16 engineered B. napus to express hpRNAs targeting the S. sclerotiorum ABHYDROLASE-3 through HIGS and demonstrated reduced disease severity in inoculated stem tissues, coupled with significant ABHYDROLASE-3 transcript reduction. As S. sclerotiorum infection commonly initiates in leaves prior to progressing to stem tissues17,18, understanding how ABHYDROLASE-3 silencing influences the host response directly at the site of infection can provide insight into the control S. sclerotiorum prior to systemic infection. Further, while ABHYDROLASE-3 silencing has demonstrated reduced disease severity against S. sclerotiorum using SIGS and HIGS technology, its role in pathogenesis is still largely unknown and identification of interacting partners and pathways may provide new candidates to target through RNAi.

Plants activate innate immunity pathways in response to pathogen attack and are dependent on the recognition of specific pathogen-derived molecules19. These pathways are commonly separated into necrotrophic and biotrophic responses and demonstrate distinct modes of pathogen detection. For example, necrotrophic pathogens are detected through the presence of pathogen and damage associated molecular patterns (PAMPs/DAMPs) via pattern recognition receptors (PRRs) in pattern-triggered immunity (PTI), while in contrast, biotrophic pathogens are detected via intracellular NBS-LRR proteins detecting secreted effector molecules in effector-triggered immunity (ETI)19. Activation of these recognition pathways induces specific defense hormones which can further activate down-stream defense processes to appropriately respond to pathogen attack. These hormones include jasmonic acid (JA) and ethylene (ET) which can directly activate the induced systemic resistance (ISR) pathway and are associated with promoting physical barriers against necrotrophic pathogens20. Other phytohormones like salicylic acid (SA) activate the systemic acquired resistance pathway (SAR) leading to a local hypersensitive response as a defense against biotrophic infection21,22. While S. sclerotiorum has been previously categorized as a necrotrophic fungal pathogen, recent reports provide evidence of an early biotrophic phase23,24. S. sclerotiorum secretes pathogenicity factors including oxalic acid (OA) and a suite of cell wall degrading enzymes to weaken and break down host cells upon infection25,26. During the initial stages of infection, a brief biotrophic phase may occur, where OA and other secreted pathogenicity factors act to suppress host defenses, specifically the SA-mediated SAR pathway23,24. The inability of S. sclerotiorum to successfully suppress SAR may lead to increased resistance against S. sclerotiorum infection. Further support for this initial biotrophic phase comes from the identification of secreted effector molecules essential for pathogenicity27 and intracellular NBS-LRRs necessary for successful defense against S. sclerotiorum28. Lastly, synergistic effects between SA-mediated SAR and JA/ET-mediated ISR have been described against necrotrophic infection29 and that removal of either of these defense hormones will result in an unsuccessful defense response30.

In the current study, we engineered A. thaliana expressing hpRNA complementary to S. sclerotiorum ABHYDROLASE-3 under the activity of a CaMV 35S promoter (AT35S::SS1G_01703RNAi herein referred to as AT1703). AT1703 A. thaliana was more tolerant to S. sclerotiorum infection with reduced lesion size, fungal load and ABHYDROLASE-3 transcript abundance compared to wild-type Col-0. Light microscopy of the infection process further identified reduced epidermal and mesophyll tissue degradation and vascular tissue colonization in AT1703 lines. Global RNA sequencing revealed enrichment of the SA-mediated SAR pathway and predictive gene regulatory network analysis predicted heat shock factor (HSFA4A, HSFA8) and TGA (TGA10) transcription factors as putative regulators of resistance in AT1703. Analysis of ABHYDROLASE-3 predicted interacting partners suggests this pathogenicity factor plays a role in the methionine salvage pathway which is necessary for successful polyamine biosynthesis. Taken together, these data demonstrate the utility of HIGS in slowing S. sclerotiorum infection and highlight the importance of the SA-mediated SAR pathway in contributing to successful defense against S. sclerotiorum.

Results

HIGS of S. sclerotiorum ABHYDROLASE-3 slows infection in transgenic AT1703 Arabidopsis

hpRNA complementary to the ABHYDROLASE-3 gene coding sequence of S. sclerotiorum (SS1G_01703) was transformed into wild-type Col-0 A. thaliana under the control of the CaMV 35S promoter and propagated to the T3 generation. Three independent insertion lines were propagated to the T2 generation (Supplementary Fig. S1a), and our top performing line (AT1703.1, herein referred to as AT1703) was selected for RNA sequencing and further analysis at the T3 generation due to its significantly reduced lesion size and ABHYDROLASE-3 transcript reduction compared to wild-type Col-0 (Supplementary Fig. S1b). Light microscopy of leaf cross sections using the chitin-targeting stain lactophenol blue showed no apparent differences in uninfected AT1703 and wild-type Col-0 leaves (Fig. 1a). However, at 2 dpi, wild-type Col-0 leaves show an increased abundance of S. sclerotiorum in both mesophyll and vascular tissues (Fig. 1b). At 3 dpi S. sclerotiorum shows further colonization of vascular tissue in wild-type Col-0 compared to AT1703 leaves, coupled with decreased levels of epidermal and mesophyll tissue degradation in AT1703 lines (Fig. 1c). At 1 dpi, no differences in lesion size or fungal abundance were observed (Fig. 1d,e), while at 2 dpi, S. sclerotiorum ABHYDROLASE-3 transcript levels showed a 73% reduction in AT1703 with no significant differences in lesion size or fungal load compared to wild-type Col-0 (Fig. 1f). However, by 3 dpi AT1703 showed a 48% reduction in lesion size, an 84% reduction in fungal load, and a 93% reduction in ABHYDROLASE-3 transcript accumulation compared to wild-type Col-0 plants (Fig. 1d,f). While this significant transcript reduction is present as early as 2 dpi, these data suggest a ~ 24-h lag phase where transcript abundance is reduced but insufficient to immediately slow infection (Fig. 1d,f).

Figure 1
figure 1

Whole-leaf and cross section of wild-type Col-0 and AT1703 A. thaliana uninfected control and infected with S. sclerotiorum at (A) 0, (B) 2, and (C) 3 days post inoculation. Five µM cross-sections generated using Leica RM2245 and stained with (A) toluidine blue and (B,C) lactophenol cotton blue. (D) lesion size (mm2), (E) fungal load and (F) target transcript reduction in Col-0 and T3 AT1703 leaves infected with S. sclerotiorum. A. thaliana inoculated with 1 × 106 µL ascospore solution and measured at 2- and 3-days post inoculation. (D) Sample size of n = 15 leaves per line were measured using ImageJ software and normalized to wild-type Col-0 leaves for lesion measurements. Error bars represent standard error and significance was determined using a Student’s t-test (p < 0.05). * = significance. (E) Fungal load quantified using relative abundance of 18S S. sclerotiorum rDNA and calculated using ∆CT against wild-type Col-0. Each line consists of three biological replicates, each containing three infected A. thaliana leaves. (F) Target transcript abundance calculated using the ∆∆CT method with Sac7 as a reference gene. Each line consists of three biological replicates, each containing three infected A. thaliana leaves for (E) and (F). Error bars represent standard error. Scale bars = 50 µm. e, epidermis; m, mesophyll; x, xylem; p, phloem; v, vasculature; s, S. sclerotiorum.

Full size image

Global RNA-sequencing reveals enriched gene activity of the salicylic acid-mediated SAR pathway in AT1703

To uncover the underlying gene expression patterns between AT1703 and wild-type Col-0 we performed global RNA sequencing on uninfected (0 dpi) leaves as well as directly at the site of infection at mid (2 dpi) and late (3 dpi) stages of S. sclerotiorum leaf infection. Hierarchical clustering of raw count expression values revealed distinct clusters forming between uninfected (0 dpi) and S. sclerotiorum infected tissues (Fig. 2). Genotypes also clustered in response to S. sclerotiorum infection compared to infection time points. Global gene activity showed no significant differences in the proportion of low, moderate or highly accumulated transcripts, with on average 20%, 21% and 58% of detected transcripts showing low, moderate and high-count levels respectively (Fig. 2). Further, total transcript detection between genotypes is similar, with an increase in 1813 and 1348 detected transcripts in AT1703 compared to Col-0 at 2 and 3 dpi respectively due to an increased abundance of S. sclerotiorum transcripts in wild-type Col-0.

Figure 2
figure 2

Global gene activity in the A. thaliana-S. sclerotiorum pathosystem. Hierarchical clustering and global gene activity of wild-type Col-0 and AT1703 at 0-, 2- and 3-days post inoculation. Number of transcripts detected across all treatments were identified through RNA sequencing. Transcripts with a count value > 1 are considered detected. Detected transcripts are subdivided into low (1 ≥ 5), moderate (5 > 25), or high (≥ 25) abundance levels.

Full size image

Differential expression analysis identified shared and specifically up-regulated gene sets in response to S. sclerotiorum in AT1703 and wild-type Col-0 (Fig. 3a). For example, 335 and 235 genes were specifically up-regulated at 2 and 3 dpi respectively, while 444 genes were up-regulated at both timepoints in AT1703. Further, 449 and 235 genes were down-regulated in AT1703 relative to Col-0 at 2 and 3 dpi respectively, and 351 genes were down-regulated at both timepoints (Supplementary Dataset S4). Next, to identify putative biological processes associated with these differentially expressed gene sets, we performed Gene ontology (GO) term enrichment (Fig. 3b). Data show the SA-mediated SAR pathway (p = 2.05 × 10–27) and SA biosynthesis (p = 1.16 × 10–31) as well as enrichment of the hypersensitive response (p = 3.07 × 10–7) in gene sets up-regulated at 2 and 3 dpi in AT1703. The enrichment of SA biosynthesis, SAR and the hypersensitive response indicate a strong role of the SAR defense pathway in successful defense response and incompatible interaction (p = 1.28 × 10–13) found in AT1703 lines. While SA biosynthesis and SAR demonstrate shared enrichment between AT1703 at 2 and 3 dpi, it is worth noting that statistically significant enrichment is also found specifically at 2 dpi, while no such enrichment is found for down-stream activated defense processes like the hypersensitive response and incompatible interaction (Fig. 3b). Further, 750 genes were up-regulated in response to S. sclerotiorum across both genotypes and timepoints (Fig. 3a). This shared gene set showed significant enrichment of PTI associated JA (p = 7.77 × 10–34) and ET biosynthesis (p = 1.32 × 10–49), induced systemic resistance (p = 6.63 × 10–8) and PAMP induction of immune response (p = 1.71 × 10–4) (Supplementary Dataset S1), suggesting that activity of these JA/ET-mediated defense pathways alone in wild-type Col-0 are insufficient to slow S. sclerotiorum infection without enrichment of SA-mediated SAR, providing additional evidence of the synergism of SA-mediated SAR and JA/ET-mediated ISR/PTI in mounting a successful defense against S. sclerotiorum. In contrast, enrichment of up-regulated genes specifically in wild-type Col-0 show association with programmed cell death and cell wall reinforcement, including leaf senescence (p = 1.97 × 10–6), autophagy (p = 8.88 × 10–6), cell wall modification (p = 2.61 × 10–6) and organization (p = 2.75 × 10–4), and lignin biosynthesis (p = 7.22 × 10–7).

Figure 3
figure 3

Identification of differentially expressed genes and enriched GO terms in response to S. sclerotiorum infection in wild-type Col-0 and AT1703. (A) Venn diagram representing up-regulated genes in infected wild-type Col-0 and AT1703 A. thaliana. (B) Heatmap of enriched GO terms associated with shared up-regulated gene patterns. GO terms considered statistically significant with a hypergeometric log10 p-value < 0.001. SH shared.

Full size image

To explore potential regulators of immunity in AT1703 expressed throughout infection, we performed a transcription factor network analysis using SeqEnrich in our AT1703 up-regulated gene set in response to S. sclerotiorum. Here, we identified the TGA transcription factor TGA10, as well as the HSFs transcription factors HSFA4A and HSFA8, which are both predicted to regulate SA-signalling and SAR activity (Fig. 4a). Further, we explored differential expression between AT1703 and wild-type Col-0 in response to S. sclerotiorum in genes that have been previously described to interact with these transcription factors. Here, we found up-regulation of the TGA interacting ENHANCED DISEASE SUSCEPTIBILITY 1 (EDS1), NONEXPRESSER OF PR GENES 1 (NPR1) and ISOCHORISMATE SYNTHASE 1 (ICS1) in AT1703 A. thaliana, as well as up-regulation of the HSFA4A interacting kinase MITOGEN-ACTIVATED PROTEIN KINASE 3 (MPK3) (Fig. 4b).

Figure 4
figure 4

Differential expression and transcription factor network analysis of up-regulated shared AT1703 gene set in response to S. sclerotiorum. (A) Predictive transcription factor network analysis in shared AT1703 shared gene set in response to S. sclerotiorum. Yellow-green rounded squares represent significantly up-regulated transcription factors identified in AT1703 gene set, pink diamonds represent transcription factor DNA binding motifs, teal circles represent GO terms, and edges between DNA motifs and GO terms represent enriched motif sequences involved in the regulation of the GO term. (B) Log2 fold change of infected AT1703 and wild-type Col-0 at 2- and 3-days post inoculation relative to respective uninfected treatments.

Full size image

ABHYDROLASE-3 silencing influences expression of polyamine synthase genes in S. sclerotiorum

While AT1703 demonstrates significant ABHYDROLASE-3 transcript reduction during S. sclerotiorum infection, we had yet to explore the influence of ABHYDROLASE-3 silencing on gene expression within the fungus itself. Three predicted interacting partners of ABHYDROLASE-3 (SS1G_02233, SS1G_14434, SS1G_03597) were identified in S. sclerotiorum using stringdb (https://string-db.org/) (Fig. 5a). While SS1G_14434 and SS1G_03597 have yet to be significantly characterized, SS1G_02233 encodes a 5′-methylthioadenosine phosphorylase (MTAP) that prevents accumulation of the polyamine biosynthesis by-product S-methyl-5′-thioadenosine (MTA), inhibiting polyamine biosynthesis and ultimately fungal growth31,32. Here, all three predicted interacting partners showed significant transcript reduction in S. sclerotiorum that was infecting AT1703 compared to wild-type Col-0 at 2 and 3 dpi (Fig. 5b). Further, reduction in these predictive partners was consistent with our target gene ABHYDROLASE-3, showing at least a 50% increase in reduction from 2 to 3 dpi. To further support the role of ABHYDROLASE-3 in polyamine biosynthesis, we quantified expression of genes encoding the rate-limiting enzyme in polyamine biosynthesis ORNITHINE DECARBOXYLASE (ODC) as well as down-stream enzymes SPERMINE SYNTHASE and SPERMIDINE SYNTHASE. Transcript reduction was observed for all three genes in S. sclerotiorum challenging AT1703, suggesting a role of ABHYDROLASE-3 in polyamine biosynthesis during S. sclerotiorum infection.

Figure 5
figure 5

Transcript abundance of ABHYDROLASE-3 (SS1G_01703) predicted interacting partners using RT-qPCR. (A) predictive interacting network of ABHYDROLASE-3 generated through Stringdb based on gene fusion and co-expression analyses. (B) RT-qPCR target transcript abundance of predicted interacting partners in S. sclerotiorum challenging AT1703 A. thaliana at 2- and 3-days post inoculation relative to wild-type Col-0. Target transcript abundance calculated using the ∆∆CT method with Sac7 as a reference gene. Each line consists of three biological replicates, each containing three infected A. thaliana leaves. Error bars represent standard error and significance was determined using a Student’s t-test (p < 0.05) with a Bonferroni correction. *Significance.

Full size image

AT1703 is not tolerant to the phytopathogen Botrytis cinerea

ABHYDROLASE-3 targeting hpRNA expressed by AT1703 was carefully designed to not share 21-nt siRNAs with transcripts of related species including the closely related necrotrophic pathogen B. cinerea. To study whether AT1703 was also tolerant to B. cinerea, we next inoculated both wild-type Col-0 and AT1703 with B. cinerea spores (Fig. 6a). Here, no significant differences in lesion size (Fig. 6b), fungal load (Fig. 6c) or target transcript abundance (Fig. 6d) were found using the B. cinerea ABHYDROLASE-3 homolog BC1G_08022, suggesting no off-target effects in B.cinerea challenging AT1703.

Figure 6
figure 6

Lesion size (mm2), fungal load and target transcript knock down in wild-type Col-0 and T3 transgenic AT1703 A. thaliana leaves infected with B. cinerea. A. thaliana inoculated with 1 × 106 µL ascospore solution and measured at 3 days post inoculation. (B) Sample size of n = 15 leaves per line were measured using ImageJ software and normalized to wild-type Col-0 leaves for lesion measurements. Error bars represent standard error and significance was determined using a Student’s t-test (p < 0.05). (C) Fungal load quantified using relative abundance of B. cinerea β–tubulin and calculated using ∆CT against wild-type Col-0. Each line consists of three biological replicates, each containing three infected A. thaliana leaves. (D) Target transcript knockdown calculated using the ∆∆CT method with β–tubulin as a reference gene. Each line consists of three biological replicates, each containing three infected A. thaliana leaves for (C) and (D). Error bars represent standard error.

Full size image

Discussion

RNAi technology has the demonstrated ability to control pathogenic fungi in host plants through constitutively expressed dsRNA molecules in transgenic plants via host-induced gene silencing (HIGS). The constitutive expression of fungal-targeting dsRNAs allows for a durable means of protection against pathogen attack throughout the host plant lifecycle16,33,34. Protection of the plant relies on successful uptake and processing of dsRNA and processed siRNAs which can be generated in the host prior to uptake by the challenging pathogen3. Cross-kingdom trafficking of both dsRNAs and processed siRNA has been demonstrated across the host–pathogen interface in necrotrophic pathogens including S. sclerotiorum and the closely related B. cinerea, making these species ideal candidates for control through RNAi35,36. Here, we engineered A. thaliana expressing hpRNA molecules complementary to the previously identified ABHYDROLASE-3 in S. sclerotiorum.

Successful S. sclerotiorum infection involves the penetration and destruction of host epidermis and mesophyll tissue followed by colonization of the vasculature28,37. Once the vascular tissue has been colonized, infection is considered systemic, being able to cut off host nutrient supply while the fungus extends mycelia conveniently throughout host tissues. Susceptible genotypes have demonstrated preferential growth throughout the vascular tissue in Brassica species compared to respective tolerant lines38. While microscopy of S. sclerotiorum infected wild-type Col-0 and AT1703 demonstrates reduced S. sclerotiorum abundance in AT1703, we also find decreased levels of epidermal and mesophyll tissue degradation and reduced abundance of vascular tissue fungal colonization, thereby preventing systemic infection throughout the plant. Limiting systemic infection has significant implications on reducing yield losses from S. sclerotiorum infection. This was demonstrated by Wytinck et al.16, where transgenic B. napus lines expressing the same ABHYDROLASE-3 hpRNA construct against S. sclerotiorum show reduced vascular tissue colonization coupled with an increase in seed mass compared to untransformed B. napus post whole-plant S. sclerotiorum infection. As infection typically initiates on mature leaves before progressing into stem tissue through the vascular system39,40, it is critical to reduce vascular colonization directly at the site of infection, which can ultimately slow S. sclerotiorum progression throughout the host.

Target gene silencing in the pathogen is essential to slow the infection through HIGS. AT1703 plants demonstrate significant ABHYDROLASE-3 transcript reductions in S. sclerotiorum at 1 and 2 dpi, while reduced lesion size and fungal load are only identified at 3 dpi relative to wild-type Col-0 plants. Wytinck et al.16 showed similar results in transgenic B. napus, with significant reductions in ABHYDROLASE-3 transcript abundance and fungal load observed at 1 and 2 dpi respectively. In contrast, prior HIGS studies targeting S. sclerotiorum essential genes (TRX1, OA) detected both target transcript reduction and reduced lesion size at 2 dpi in host plants33,41, suggesting this delay to be specific to HIGS plants targeting ABHYDROLASE-3. These findings may be attributed to down-stream interactions of ABHYDROLASE-3 with other interacting partners. Reduced activity of ABHYDROLASE-3 in addition to regulators of polyamine biosynthesis are likely responsible for or at least contribute to the inhibition of fungal growth. However, further characterization of ABHYDROLASE-3 interacting partners and the polyamine biosynthesis pathway during ABHYDROLASE-3 silencing are required to better understand this delay.

While the utility of HIGS in controlling S. sclerotiorum has been demonstrated in multiple host plants16,33,34, the underlying host response of these transgenic plants, as well as host–pathogen interactions in this system are less understood and dependent upon the specific gene being silenced. Here, global RNA sequencing and differential expression analyses provide insight into the response of AT1703 leaves to S. sclerotiorum directly at the site of infection. Moreover, the mode of pathogenicity of S. sclerotiorum has been recently debated, as emerging studies have suggested the presence of a brief biotrophic stage during early infection highlighted by the suppression of the SA-mediated SAR pathway23,24. SA biosynthesis in response to pathogen attack occurs through the isochorismate synthase (ICS) pathway, where ICS1 is essential for SA biosynthesis and successful SAR42. Downstream of SA accumulation, NPR1 activity is required for expression of pathogenesis-related (PR) genes involved in the SAR response through SA signal transduction42,43. AT1703 shows significant enrichment of the SA-mediated SAR pathway compared to wild-type Col-0, with enrichment of both ICS1 and NPR1 in response to S. sclerotiorum infection, suggesting an inability to suppress SAR while infecting AT1703 A. thaliana. Further, while NPR1 does not contain a DNA binding domain, it acts as a transcriptional cofactor, enhancing TGA transcription factor binding affinity to PR genes upon translocation into the nucleus in response to SA44. While multiple TGA transcription factors have been shown to interact with NPR1 in response to SA in A. thaliana45, TGA10 interaction has yet to be explored. Here, TGA10 was identified as the lone member of the TGA transcription factor family significantly enriched in AT1703 A. thaliana in response to S. sclerotiorum and was identified in our predictive transcription factor network to regulate SAR activity. Further, TGA10 has been demonstrated to be necessary for activation of reactive oxygen species (ROS) response in PTI in response to the bacterial PAMP flg22, where tga10 mutants were unable to activate flg22-response genes and showed reduced ROS production46, highlighting its importance in the activation of defense genes in response to pathogen attack. Our network analysis further identified two HSF transcription factors (HSFA4A, HSFA8) associated with the SA signalling pathway that are up-regulated in AT1703 in response to S. sclerotiorum. While HSFs have been shown to interact directly with EDS1, resulting in SA accumulation in response to pathogen attack47,48, similarly to TGA10, HSFA4A and HSFA8 have specifically been implicated in the host ROS response as well as activation of the host antioxidant system49,50. Activation of these HSFs can act through phosphorylation by MPK3 in response to stress, which also demonstrates significant up-regulation in AT1703 A. thaliana in response to S. sclerotiorum49,51,52.

While these data suggest SAR activity is an essential component of a successful defense response against S. sclerotiorum infection, SAR alone is insufficient to protect the host, as quantitative disease resistance is complex and requires the coordination of multiple defense pathways and processes53. For example, SA-mediated SAR and JA/ET regulated ISR pathways exhibit synergistic effects in defense against necrotrophic pathogen attack29. This was demonstrated using npr1 knockout Arabidopsis mutants deficient in SA signalling, coi1 mutants deficient in JA signalling, and ein2 deficient in ET signalling, which all show hyper-susceptibility phenotypes due to the inactivity of SAR and ISR pathways30. Differential expression analysis not only identified enrichment of SA-mediated SAR in AT1703, but also JA/ET biosynthesis and the ISR pathway which were significantly enriched in both AT1703 and wild-type Col-0 lines relative to their respective uninfected leaves, suggesting both defense pathways are essential in mounting a successful defense response against S. sclerotiorum.

To better understand the different defense responses to infection by wild-type Col-0 and AT1703 plants, we further explored defense processes enriched specifically in wild-type Col-0 and not AT1703. S. sclerotiorum demonstrates necrotrophic growth, especially at later stages of infection. Thus, the bi-products of gene activity belonging to autophagy and leaf senescence found in wild-type Col-0 can serve as nutrients for the progressing pathogen to continue growth as it feeds on necrotic host tissue54,55. With increased necrotic host tissue available to S. sclerotiorum in wild-type Col-0, host plants would require mechanical defenses through cell wall reinforcement and lignification in response to cell wall degrading enzymes being consistently secreted during actively progressing infection56. Lignification in response to S. sclerotiorum has previously been described by Höch et al.56, where genes associated with lignin biosynthesis showed significantly increased expression levels in susceptible B. napus lines compared to moderately resistant lines, specifically during mid to late stages of infection. These findings agree with our data where we find enrichment of lignin biosynthesis and cell wall reinforcement in wild-type Col-0 lines compared to AT1703 throughout S. sclerotiorum infection.

While we have examined the influence ABHYDROLASE-3 silencing has on the host response, there is also a need to explore differences within the pathogen and the effect that ABHYDROLASE-3 silencing has at the molecular level. ABHYDROLASE-3 is annotated as a tRNA synthetase but also predicted to be involved in aflatoxin biosynthesis. However, this gene has yet to be further characterized beyond its ability to slow S. sclerotiorum infection through SIGS and HIGS RNAi treatments15,16. Using the stringdb (https://string-db.org/) protein–protein interaction network database, we identified three predictive interacting partners of ABHYDROLASE-3 (SS1G_14434, SS1G_03597 and SS1G_02233). Expression levels of ABHYDROLASE-3 and these predicted interacting partners in S. sclerotiorum were coordinately reduced in transgenic AT1703 compared to wild-type Col-0. While SS1G_14434 and SS1G_03597 have yet to be functionally characterized in S. sclerotiorum, SS1G_02233 encodes MTAP, which plays a critical role in cleaving and processing accumulated MTA, a by-product of polyamine biosynthesis31,32. Buildup of MTA through inactivity of MTAP ultimately results in the inhibition of polyamine biosynthesis31,32. This in turn can have significant impacts on fungal growth, as the three most widely distributed polyamines, putrescine, spermidine and spermine have been shown to be necessary for successful growth and cell differentiation32,57. The ornithine decarboxylase (SS1G_05468) is an essential enzyme of polyamine biosynthesis, converting ornithine into the polyamine putrescine32,57. Once ornithine is converted into putrescine, there are further down-stream synthase genes necessary for conversion into both spermine and spermidine. These enzymes include spermine synthase (SS1G_12906) to convert putrescine into spermine, and spermidine synthase (SS1G_10282) to convert spermine into spermidine. Our RT-qPCR analysis of these essential polyamine synthesis genes shows transcript reduction during S. sclerotiorum infection of AT1703 plants compared to their wild type susceptible counterparts, suggesting a potential interaction between ABHYDROLASE-3 and the identified predictive partners of the polyamine biosynthesis pathway. Further, as putrescine can also be produced through spermine and spermidine catabolism, expression of rate limiting enzymes SSAT and PAO can also be explored to improve our understanding of ABHYDROLASE-3 influence on polyamine metabolism32. While the potential direct or indirect interaction of ABHYDROLASE-3 and these predictive interacting partners is still unclear, these data provide some preliminary insights of the role ABHYDROLASE-3 plays in S. sclerotiorum pathogenicity and further identifies a group of promising RNAi candidate genes through the polyamine synthesis pathway.

One benefit of RNAi and HIGS technology in crop protection is the ability to design pathogen specific molecules that are ineffective against potentially beneficial species in the ecosystem4,58. This requires the careful design of target sequences within genes, as dsRNA molecules containing as few as three overlapping 21-mer nucleic acid sequences have demonstrated the ability to significantly reduce off-target transcripts59. B. cinerea serves as an ideal candidate to study these potential off-target effects in controlling S. sclerotiorum given both pathogens share similar hosts and disease cycles in addition to having a high level of sequence homology60,61. McLoughlin et al.15 showed that dsRNA targeting the ABHYDROLASE-3 B. cinerea homolog BC1G_08022 was able to reduce B. cinerea lesion size, fungal load and BC1G_08022 transcript abundance when applied to B. napus leaves prior to B. cinerea inoculation. While BC1G_08022 is capable of slowing B. cinerea infection when targeted, our dsRNA region targeting S. sclerotiorum ABHYDROLASE-3 shows no 21-mer overlapping regions with off-target species (Supplementary Dataset S2) and AT1703 A. thaliana challenged with the closely related necrotrophic fungal pathogen B. cinerea shows no significant differences in lesion size, fungal load or BC1G_08022 transcript reduction.

Taken together, these data demonstrate the efficacy of HIGS as an alternative control measure in slowing S. sclerotiorum infection through targeted silencing of S. sclerotiorum ABHYDROLASE-3. Reduction of ABHYDROLASE-3 in S. sclerotiorum provides A. thaliana the ability to increase innate defense responses through SA-mediated SAR at the global mRNA level. The identification of co-silenced ABHYDROLASE-3 interacting partners further uncovers a promising group of candidates for RNAi control against fungal pathogens. Finally, with ongoing concerns of potential off-target effects in RNA-based applications, we demonstrate the specificity of HIGS targeting S. sclerotiorum in AT1703 to a closely related fungus. Taken together, we provide evidence that HIGS applications can be used as an effective and sustainable strategy in crop improvement against necrotrophic pathogens.

Methods

Generation of transgenic Arabidopsis

DNA constructs consisted of a CaMV-35S promoter driving expression of a sense and antisense SS1G_01703 gene fragment separated by an intron region to form hpRNA post transcription. Cloning of this construct into Agrobacterium tumefaciens followed the Gateway cloning protocol (Invitrogen, Carlsbad, CA, US) using primers found in Supplementary Dataset S3 and following the methods of16. Target gene sequences were amplified using Phusion Taq (Thermo Scientific, Waltham, MA, US) under the following conditions: 98 °C for 30 s; 35 cycles of: 98 °C for 10 s, 57 °C for 30 s, and 72 °C for 30 s; and a final extension of 72 °C for 5 min. Amplicons were gel purified (New England Biolabs, Ipswich, MA, US) and digested using FastDigest KpnI and XhoI (Thermo Scientific, Waltham, MA, US) according to the manufacturer’s protocols. Initially, gene fragments were ligated into the pENTR4 vector before insertion into the pHellsgate8 destination vector. To confirm insert identity, entry vectors were sequenced at the Centre for Applied Genomics in Toronto, Ontario. Inserts were then recombined into the destination vector using a Gateway LR clonase reaction (ThermoFisher Scientific) following the manufacturer’s instructions, with the modification of adding 4:1 entry vector to destination vector ratio according to Wytinck et al.16. A cold shock treatment was used to transform Agrobacterium. Successful transformants were selected using colony PCR with XhoI and XbaI separate restriction enzyme digestions. Cultures of Agrobacterium were grown to OD 1.6–2.0, pelleted using centrifugation, and re-suspended in MS media and 0.001% Silwet L-77. To transform A. thaliana, seeds were initially sterilized using alternating washes of 70% and 95% ethanol then placed on Murashige and Skoog (MS) media (Sigma Aldrich). Seeds were vernalized for three days at 4 °C prior to being grown in a constant light incubator. Upon formation of first leaves A. thaliana seedlings were transplanted into Sunshine Mix #1 and grown to maturity at 22 °C. Mature flowering A. thaliana plants were dipped into the Agrobacterium culture and kept in high humidity conditions for two days. Floral dips were repeated five times before seeds were dried and harvested. Transformants were selected using MS and 150 μg/mL kanamycin media62. PCR was used to confirm the presence of the S. sclerotiorum gene using the same primers used for cloning within the plant (Supplementary Dataset S3). Three confirmed independently transformed A. thaliana lines were grown to the T2 generation and tested for their ability to silence ABHYDROLASE-3 transcripts and slow S. sclerotiorum infection. The top performing line (AT1703.1) was subsequently selected for further RNA sequencing analysis.

Arabidopsis thaliana infection assays

Sclerotinia sclerotiorum ascospores were collected at the Morden Research and Development Centre, Agriculture and Agri-Food Canada, Morden, MB, Canada and stored at 4 °C in desiccant in the dark according to Ref.15. S. sclerotiorum ascospore inoculum was made by suspending a 1 × 106 spore/mL concentration of spores in a potato dextrose broth (PDB) and peptone solution (24 g PDB, 10 g peptone, 1 L water). Using a pipette, 10 µL of the solution was transferred onto mature transgenic and wild-type Col-0 A. thaliana leaves at a n = 15 leaves per treatment. Plants were stored in growth chambers at room temperature (21.0 °C), with plant trays being sealed with lids to maintain high levels of humidity for three days, allowing for infection progression in planta. At 2- and 3-days post inoculation (dpi), lesion area was quantified using ImageJ software and leaves were collected for RNA sequencing, fungal load, and transcript abundance experiments. Fifteen infection sites were quantified per treatment while three leaves were collected per bio-replicate with three biological replicates collected per treatment. Botrytis cinerea infections were performed using ascospores from the Saint-Jean sur Richelieu Research and Development Centre, Agriculture and Agri-Food Canada, QC, Canada following the same protocol for S. sclerotiorum ascospore inoculations.

DNA extraction and plant genotyping

Two-week-old A. thaliana leaves were collected and ground using a GenoGrinder 2000 (Spex CertiPrep, Metuchen, New Jersey, USA). DNA was extracted according to Wytinck et al.16 with DNA extraction buffer [0.25 M NaCl, 1 M 2-amino-2-(hydroxymethyl)-1,3-propanediol (TRIS)-HCl pH 7.5, 25 mM EDTA pH 8, 0.5% SDS]. DNA was precipitated with isopropanol and subsequently washed with 75% ethanol before being dissolved in water. PCR was performed using GoTaq® Green Master Mix (Promega, Madison WI, USA) according to manufacturer’s instructions using ABHYDROLASE-3 specific primers (Supplementary Dataset S3). Thermocycler conditions were 94 °C for 2 min. followed by 35 cycles of: 94 °C for 1 min, 56 °C for 30 s, 72 °C for 35 s. and a final extension at 72 °C for 5 min. PCR products were then loaded on a 1% agarose gel containing ethidium bromide with 2 μL of sample loaded per well against FastRuler Middle Range DNA Ladder (Thermo Fisher) as a reference. Gels were visualized under UV light using the Axygen® Gel Documentation System (Axygen, Corning, NY, US).

Lactophenol cotton blue stain microscopy

Arabidopsis thaliana leaf tissue was embedded in historesin (Leica, Wetzlar, GER) and vacuum infiltrated in 2.5% glutaraldehyde and 1.6% paraformaldehyde fixative solution. Methylcellosolve (Sigma-Aldrich, St. Louis, MO, USA) was used to remove pigment followed by three transfers of 100% ethanol, one every 24 h to dehydrate tissue. Tissue was then infiltrated with historesin during a 3-day period of increasing concentrations of resin. Day 1: 1:3 historesin:100% alcohol, day 2: 2:3 historesin:100% alcohol, and day 3: 100% historesin. Tissue was then embedded in blocks with an embedding solution of activated historesin, hardener and polyethylene glycol according to Wytinck et al.16. Sections were cut at 5 μM with a Leica RM2245 microtome and S. sclerotiorum infected tissues stained with lactophenol cotton blue for 20 min, while uninfected 0 days post inoculated tissues were stained with 0.1% toluidine blue O for 15 min. Uninfected and infected wild-type Col-0 and AT1703 A. thaliana leaves were imaged using a Leica DFC450C camera.

RT-qPCR analysis

Sclerotinia sclerotiorum and B. cinerea RT-qPCR primers were designed using Primer3 (http://bioinfo.ut.ee/primer3/) and subsequently validated using the Primer BLAST tool (www.ncbi.nlm.nih.gov/tools/primer-blast) with a product range of 70–90 bp, GC% range of 40–60% and a primer size range of 18–23 bp. Transcript abundance was measured on the Bio-Rad CFX96 Connect Real-Time system using SsoFast EvaGreen Supermix (Bio-rad Laboratories, Hercules, CA, US) in 10 µL reactions according to manufacturer’s protocol using the following conditions: 95 °C for 30 s, and 45 cycles of: 95 °C for 2 s and 60 °C for 5 s. Melt curves were performed at a range of 65–95 °C with 0.5 °C increments to assess nonspecific amplification and primer dimers. Relative transcript accumulation was calculated using the ΔΔCt method63, relative to Sac7 (SS1G_12350) for S. sclerotiorum and tubA (BC1G_00122) for B. cinerea using three biological replicates per treatment and three technical replicates per biological replicate. The same ΔΔCt method was used to quantify expression levels of RNAi machinery in S. sclerotiorum and have been used and validated in Refs.15,16. All RT-qPCR primers used in this study can be found in Supplementary Dataset S3.

mRNA library preparation and RNA sequencing

RNA was isolated using the Ambion® RNaqueous® micro kit (ThermoFisher Scientific) according to manufacturer’s instructions. RNA quality was analyzed using the Agilent 6000 Pico LabChip® and Agilent 2100 Bioanalyzer software (Agilent Technologies, USA). RNA quantity was determined using the Quant-iT™ RiboGreen® kit (ThermoFisher Scientific) according to manufacturer’s instructions. NEBNext Oligo d(T)25 beads (New England BioLabs, Ipswich, MA, USA) were used for mRNA purification. cDNA library synthesis was performed according to Ziegler et al.64 with three biological replicates per treatment each containing three A. thaliana leaves. Subsequently, 100 bp paired-end RNA sequencing was performed on the Illumina HiSeq4000 platform (Génome Québec Innovation Centre, McGill University, Montreal, Canada).

RNA sequencing analysis

Raw and processed sequence reads were deposited at the Gene Expression Omnibus under the accession GSE217513. Read quality control and adapter sequence removal was performed using Trimmomatic 0.3665 (HEADCROP:9 LEADING:30 TRAILING:30 SLIDINGWINDOW:4:30 MINLEN:50). Surviving reads were aligned to the TAIR10 A. thaliana genome using HISAT266 alignment software. Raw count data was generated using featureCounts67 and inputted into pvclust (https://cran.r-project.org/web/packages/pvclust/index.html) for hierarchical clustering analysis with a detection cutoff ≥ 1. Detected transcripts were subsequently sorted into low (1 ≥ 5), moderate (5 > 25), or high (≥ 25) abundance levels. Further, raw count data was processed in DESeq2 for differential expression analysis. Here, differentially expressed genes (DEGs) were identified between pairwise comparisons at a p-value cutoff of p < 0.0001. GO enrichment of identified DEGs was performed using SeqEnrich68 with terms being considered significantly enriched at p ≤ 0.001 according to software specifications. GO heatmap visualization was performed using the conditional formatting function in Excel. GO summary tables can be found in Supplementary Dataset S1. Predictive transcription factor networks were generated using SeqEnrich (Supplementary Dataset S5) with DEG gene lists being used as input. Networks were subsequently visualized using Cytoscape 3.9.1. software (https://cytoscape.org/).