First, we performed S. sclerotiorum infection assays of the C. sativa cola to better understand disease progression over time (Fig. 1). At one day post inoculation (dpi) no external disease symptoms were visible (Fig. 1A, i-iii). At 3 dpi, we first observed floral tissue necrosis at the site of inoculation (Fig. 1A, iv). By 5 dpi, necrosis was observed throughout the inoculated inflorescence and had extended to the inflorescence axis, nearing the main stem of the cola (Fig. 1A, v). Finally, at 7 dpi, necrosis had become widespread, affecting neighbouring inflorescences of the cola and had extended down into the main stem axis (Fig. 1A, vi). Infected necrotized tissues were pale brown in colour, and friable. Necrotic tissue present in the interior of the cola was soft and water-soaked, while necrotic tissue found towards the exterior of the cola was dry and brittle. Alignment of RNA sequencing reads to S. sclerotiorum increased in infected samples as infection time progressed, while the opposite trend was observed for reads aligned to the C. sativa genome (Fig. 1B). This finding was further supported by qPCR results quantifying relative fungal load (Supplementary Figure S1). Targeting S. sclerotiorum 18 S rDNA, data revealed S. sclerotiorum became more abundant in the C. sativa cola during the seven day infection process.

Fig. 1

S. sclerotiorum infection of the C. sativa cola. (A) Symptom progression of S. sclerotiorum infection of the C. sativa cola up to seven days post inoculation (dpi). Whole cola (i) pictured next to trimmed cola (ii), both at time zero. Trimmed infected colas pictured at 1-, 3-, 5-, and 7 dpi indicated as iii, iv, v, and vi, respectively. Scale bar = 1 cm. (B) Percent alignment of RNA sequencing reads from infected and uninfected control samples to both the C. sativa and S. sclerotiorum genomes. Error bars correspond to standard error. (C) Dendrogram of C. sativa samples based off hierarchical clustering of the top 10,000 most variable genes. Height corresponds to Euclidean distance between clusters. INF = infected, UTC = untreated control.

Global shifts in gene expression were observed in both the host plant and fungal pathogen as S. sclerotiorum initiated infection in C. sativa. Hierarchical clustering the top 10,000 most variably expressed C. sativa genes revealed that treatments clustered together based on infection status with the exception of the 1 dpi timepoint which remained clustered with uninfected samples (Fig. 1C). These results were supported by principal component analysis (PCA) of individual samples which revealed that the largest source of variation in our data was attributed to infection status (Supplementary Figure S2). Furthermore, infected samples clustered into distinct groups based on time post inoculation, highlighting large shifts in gene activity as infection progressed across the seven-day infection period. Similarly, hierarchical clustering of the top 1000 most variably expressed S. sclerotiorum genes revealed the 1 dpi timepoint to cluster with in vitro grown S. sclerotiorum, while all other infection timepoints clustered distinctly (Supplementary Figure S3). PCA of individual samples directly supported hierarchical clustering results (Supplementary Figure S4). We validated the RNA sequencing results by comparing the relative expression of the SAR marker gene PATHOGENESIS RELATED PROTEIN 1 (PR-1) using RT-qPCR. Data show PR1 accumulates at similar levels regardless of the method used to evaluate its activity (Supplementary Figure S5).

To better understand the interaction between S. sclerotiorum and C. sativa we tracked fungal infection of the cola at the cellular level directly from the site of inoculation (Fig. 2A). At 3 dpi, the inoculation site was clearly visible, as was the extension of fungal hyphae as S. sclerotiorum began to infect host floral tissue. At this timepoint, sectioning of reduced leaves proximal to the inoculation site revealed the presence of fungal hyphae along the surface of the epidermis as well as within epidermal cells, palisade mesophyll and general parenchymatic tissues, and phloem tissue of the vascular bundle (Fig. 2B). Xylem tissues remained relatively untouched whereas the phloem showed extensive colonization by the fungus as compared to uninfected reduced leaves (Fig. 2B-D). While the presence of hyphae was found throughout the reduced leaf, minimal plant cell wall degradation was visible. By 7 dpi, the C. sativa reduced leaf showed severe degradation of all tissue layers apart from the xylem (Figs. 3E). Although still structurally intact, S. sclerotiorum hyphae were visible throughout the xylem at this timepoint. The C. sativa stalked glandular trichomes of the inflorescence were also infected at this timepoint (Fig. 2F).

Fig. 2

S. sclerotiorum infection of the C. sativa inflorescence and infection progression in reduced leaves. (A) Longitudinal section of the S. sclerotiorum inoculation site (s, white arrow) three days post inoculation. Trichomes (tr) are visible throughout the inflorescence. Scale bar = 100 μm. (B) Longitudinal section of the reduced leaf within the inflorescence. S. sclerotiorum hyphae (s, white arrow) found in epidermis (epi), parenchyma (par), and vascular bundle (vb) three days post inoculation. Scale bar = 50 μm. (C) Cross section of uninfected reduced leaf. Vascular bundle (vb), epidermal cells (epi) and parenchyma (par). Scale bar = 50 μm. (D) Cross section of reduced leaf three days post inoculation. S. sclerotiorum hyphae (s, white arrow) present throughout vascular bundle and parenchyma (par). Tracheary cells of the xylem (x) remain intact. Scale bar = 50 μm. (E) Cross section of reduced leaf seven days post inoculation. S. sclerotiorum hyphae (s, white arrow) present in epidermis (epi), parenchyma (par), and vascular bundle. Xylem (x) tracheary cells remain intact. Scale bar = 50 μm. (F) S. sclerotiorum infection of the stalk (st) and disk cells (dc) of a glandular trichome seven days post inoculation. Secretory cavity (sc) and cuticle are still intact. Scale bar = 50 μm.

Fig. 3

Longitudinal sections of the C. sativa inflorescence stem uninfected versus seven days post inoculation with S. sclerotiorum. (A) Longitudinal section of an uninfected C. sativa inflorescence stem. Epidermis (epi), cortex (co), phloem (ph), xylem (x), and pith (p) are labelled accordingly. Scale bar = 100 μm. (B) Longitudinal section of a C. sativa inflorescence stem infected with S. sclerotiorum seven days post inoculation. Extensive tissue degradation is apparent across the epidermis (epi), cortex (co), phloem (ph), and pith (p). The xylem (x) remains relatively intact when compared to other tissue layers. S. sclerotiorum (s) is present throughout the tissue layers of the stem. Scale bar = 100 μm.

In the floral stem of the plant, S. sclerotiorum infection resulted in extensive host tissue degradation (Fig. 3). The floral stem of C. sativa is divided into distinct cell and tissue layers including the epidermis, cortex, phloem, xylem and pith (Fig. 3A). Extensive tissue degradation was revealed in infected floral stem tissues where S. sclerotiorum hyphae were abundant in the epidermis, cortex, vascular tissues, and pith of the C. sativa stem at 7 dpi (Fig. 3B). Most heavily degraded was the phloem, where cell walls had collapsed entirely, leading to the occurrence of regions of open space in areas where the fungus had used the phloem to travel further into the main stem of the plant. Although hyphae were also present within the xylem tracheids; degradation of this tissue layer was limited.

Next, we carried out differential gene expression analysis to better understand how C. sativa responds to S. sclerotiorum at the mRNA level (Fig. 4). The largest number of up-regulated differentially expressed genes (DEGs) in C. sativa were found at the intersection of 3-, 5- and 7 dpi, and of 3- and 5 dpi, with 2937 and 1855 genes, respectively (Fig. 4A). Specific to each timepoint, 4 genes were upregulated at 1 dpi, 393 genes at 3 dpi, 861 genes at 5 dpi, and 850 genes at 7 dpi. To better understand the biological and molecular processes associated with these gene sets, we conducted a GO enrichment analysis (Fig. 4B). Enriched at all timepoints were GO terms associated with terpene biosynthesis (terpene synthase activity and diterpenoid biosynthetic process), plant stress responses (abscisic acid binding and response to oxidative stress), plant defense, and chitinase activity. Specific to 3- and 5 dpi, we observed enrichment of GO terms associated with protein synthesis/transport (translation and endoplasmic reticulum to Golgi vesicle-mediated transport). Shared between 5- and 7 dpi, were GO terms involved in hormone signalling (regulation of jasmonic acid signalling and regulation of SA biosynthesis), response to wounding, and calcium/calmodulin signalling. More generally, shared between 3-, 5-, and 7 dpi were terms pertaining to oxidative stress responses (glutathione metabolic process and hypersensitive response), ethylene signalling, and protein kinase and protein ser/thr kinase activity. In our data, we identified a larger number of enriched GO terms shared between infection time points while few GO terms were enriched at specific timepoints. All GO enrichment and differential gene expression analysis results and associated P-values are provided in the supporting information (Supplementary Material 1).

Fig. 4

Upregulated differential gene expression of C. sativa infected with S. sclerotiorum over a seven-day infection period. (A) Venn diagram of significantly upregulated differentially expressed gene sets (FDR < 0.05) in response to infection. (B) Heatmap of significantly enriched GO terms (FDR < 0.01) resulting from timepoint-specific and shared subsets. A brighter yellow colour indicates greater statistical significance. dpi = days post inoculation.

In response to infection, we uncovered genes encoding chitinases and endochitinases to be upregulated in C. sativa (Fig. 5). Those with highest fold changes in response to infection at 5 dpi included ENDOCHINTINASE 2 (LOC115705823; 1982-fold) and ENDOCHITINASE-LIKE (LOC133037227; 1074-fold). Also seen to be highly upregulated at 5 dpi were genes involved the plant SAR response, including PATHOGENESIS RELATED (PR) genes. These genes included THAUMATIN-LIKE PROTEIN 1B (PR-5; LOC115710654; 3487-fold), PATHOGENESIS-RELATED PROTEIN STH-2 (PR-10a; LOC115722015; 917-fold) and MAJOR ALLERGEN PRU AV 1 (PR-10; LOC115722031; 899-fold). Additionally, we observed the induction of genes involved in JA/ET hormone signalling. JASMONATE-ZIM DOMAIN (JAZ) protein genes JAZ5 and JAZ8 in addition to ETHYLENE RESPONSIVE FACTORs (ERF) ERF096 and ERF098 were also significantly differentially expressed. Furthermore, we uncovered the notable upregulation of numerous RLK genes that included WALL ASSOCIATED RECEPTOR KINASE 2 (WAK2; LOC115708008) and WALL ASSOCIATED RECEPTOR KINASE-LIKE 1 (WAKL1; LOC115696698) which both exhibited a 500-fold increase in expression, in addition to various other serine/threonine RLKs that include G-TYPE LECTIN S-RECEPTOR-LIKE SERINE/THREONINE-PROTEIN KINASE 3 (LECRK3; LOC113032648) and LECRK4 (LOC115721224). Finally, a number of peroxidases in C. sativa were also found to be highly upregulated in response to S. sclerotiorum. These peroxidases included PEROXIDASE 57 (PER57; LOC115722259), CATIONIC PEROXIDASE 1 (LOC115720664), LIGNIN-FORMING ANIONIC PEROXIDASE (LOC115723064), and PEROXIDASE 5-LIKE (LOC115723295) with fold changes at 5 dpi of 9710, 2868, 2375, and 1470, respectively.

Fig. 5

Heatmap of significantly differentially expressed C. sativa genes (FDR < 0.05) belonging to enriched GO terms following inoculation with S. sclerotiorum. Brighter yellow colour indicates a greater fold change in expression compared to uninfected plants. dpi = days post inoculation.

We were next interested in understanding the genes and biological processes that were downregulated in C. sativa in response to infection with S. sclerotiorum (Supplementary Figure S6). The largest number of down-regulated genes were shared between 3-, 5- and 7 dpi (2866 genes; Supplementary Figure S6A). Large numbers of shared DEGs were also observed between 3- and 5 dpi (1458 genes) and 5- and 7 dpi (1179 genes). Specific to each timepoint, we identified 850 DEGs at 7 dpi, 1540 DEGs at 5 dpi, 512 at 3 dpi, and interestingly, 0 at 1 dpi. GO enrichment revealed enrichment of terms associated with photosynthesis, cellular development, and terpene synthesis (diterpenoid biosynthetic process and terpene synthase activity) were shared between 3-, 5-, and 7 dpi (Supplementary Figure S6B). Abscisic acid biosynthesis was enriched in gene sets shared between 3- and 5- dpi while biological processes associated with hormone activity like jasmonic acid biosynthesis and cytokinin signalling were specific to gene sets at 3 dpi.

While terpene/diterpenoid biosynthesis GO terms were enriched in both up and downregulated gene sets, the specific genes involved in either GO term were unique (Fig. 6). Upregulated in response to S. sclerotiorum infection were C. sativa terpene synthases that included MONOTERPENE SYNTHASE MTS1 (LOC133031472, LOC115723097, LOC133030985, LOC115723096, LOC115723095) and (-)-GERMACRENE D SYNTHASE-LIKE (LOC115707304). Both genes were very highly upregulated with fold changes seen as high as 6361 at 7dpi (LOC13303985) and 6400 at 5dpi (LOC115707304). Conversely, genes downregulated in response to infection included ALPHA-HUMULENE SYNTHASE and -SYNTHASE-LIKE (LOC115695864, LOC115695866, LOC115725506 and LOC133038934, LOC133039417, LOC115715212), (E-E)-GERANYLLINALOOL SYNTHASE (LOC115696242), (-)-LIMONENE SYNTHASE (LOC115716064, LOC115716066, LOC133037760) and MYRCENE SYNTHASE (LOC115716405, LOC133029092, LOC133037756). Some of the most downregulated genes included (E-E)-GERANYLLINALOOL SYNTHASE (14.8-fold compared to uninfected plants at 7 dpi) and ALPHA-HUMULENE SYNTHASE (LOC115725506; 14.6-fold compared to uninfected plants at 5 dpi).

Fig. 6

Heatmap of significantly up- and downregulated C. sativa genes of the diterpenoid biosynthetic process and terpene synthase activity in response to S. sclerotiorum (FDR < 0.05). Colour corresponds to log₂fold change, where a saturated orange colour corresponds to a greater log₂fold change downregulation, and a saturated blue colour corresponds to a greater log₂fold change upregulation. dpi = days post inoculation.

To investigate changes in S. sclerotiorum gene activity during C. sativa infection, we carried out differential gene expression and gene ontology (GO) term enrichment analysis (Fig. 7). Differential expression analysis revealed a high degree of shared differentially expressed genes at all sample timepoints (770 upregulated DEGs; Fig. 7A). Further, we observed the specific upregulation of 308 DEGs at 1 dpi, 95 at 3 dpi, 198 at 5 dpi, and 877 at 7 dpi. GO terms associated with host plant cell wall breakdown and carbohydrate metabolism (carbohydrate metabolic process, cellulose binding, xylan catabolic process and polygalacturonase activity) in addition to protein serine/threonine kinase activity were shared across all time points. GO terms associated with carbohydrate and cell wall breakdown, together with fungal growth and development within the host were enriched in gene sets shared between 3-, 5-, and 7- dpi. Redox processes and homeostasis were enriched in both the 3-, 5-, 7 dpi shared group, and the 3-, 7 dpi shared group (Fig. 7B).

Fig. 7

Upregulated differential gene expression of S. sclerotiorum infecting C. sativa across a seven-day infection period. (A) Venn diagram of significantly upregulated differentially expressed gene sets (FDR < 0.05) in response to infection. (B) Heatmap of significantly enriched GO terms (FDR < 0.01) resulting from timepoint-specific and shared subsets. A brighter yellow colour indicates greater statistical significance. Differentially expressed genes were compared to in vitro grown S. sclerotiorum. dpi = days post inoculation.

Differential gene expression analysis of downregulated genes identified large numbers of shared DEGs between all sampled time points (610 genes) and shared between 3-, 5- and 7 dpi (710 genes; Supplementary Figure S7). Gene ontology terms enriched across all sampled timepoints identified enriched GO terms associated with translation and protein folding (ribosome, translation, unfolded protein binding, and protein folding) in addition to mitochondrial activity (mitochondrion and tricarboxylic acid cycle).

To validate the findings of our RNA sequencing experiments, we studied the relative abundance of four C. sativa defense markers including PR-1, PR-10a, PR-5, and ENDOCHITINASE-2 using reverse-transcriptase quantitative PCR (RT-qPCR). Data show similar levels of gene activity of the four selected transcripts in both the RNA sequencing counts data and relative mRNA abundance observed using RT-qPCR (Figure S8).

Female C. sativa plants, cultivar ‘Kona’, were sourced from Rogue Botanical, a licensed grower in southern Manitoba, Canada. Plants were obtained in vegetative growth, 20 days after being clonally propagated. At 22 days old, plants were transplanted from 4-inch pots to 6-inch pots in Sunshine growing mix #4 (Sungro, Agawan, MA, USA). Plants were grown in a controlled environment chamber under long day conditions (18 h light, 6 h dark), 500 µmol/m²/s^− 1, 23 °C and 50% relative humidity. After 30 days in vegetative growth, the photoperiod was adjusted to 12 h light, 12 h dark to promote flowering. Plants were fertilized using Advanced Nutrients Sensi Grow/Bloom nutrient packages, as per manufacturer’s instructions (Advanced Nutrients, West Hollywood, CA, USA).

Experimental research on C. sativa, including the collection of plant material, were conducted in accordance with relevant institutional, national, and international guidelines and legislation.

S. sclerotiorum was grown in vitro on potato dextrose agar (BD Difco) plates supplemented with 15 µg/mL tetracycline HCl. S. sclerotiorum mycelial plugs were taken from the leading edge of a 3-day-old actively growing plate using a P1000 pipette tip. Mycelial plugs were carefully placed at the inflorescence node of the third-most distal inflorescence of the C. sativa cola using forceps. Infection took place over a seven-day period with colas being harvested after 1-, 3-, 5- and 7- days post inoculation (dpi). Both infected and untreated control (UTC) colas were harvested at each timepoint. Twelve C. sativa plants were used for this experiment with four biological replicates sampled across each timepoint.

Harvested colas were immediately trimmed down to the main floral stem, while maintaining ~ 1 cm³ of floral tissue of the inoculated inflorescence and immediately flash frozen using liquid nitrogen. Tissue was ground to a fine powder using a mortar and pestle with liquid nitrogen prior to RNA extraction.

RNA was extracted using the Purelink Plant RNA Reagent (Invitrogen, Waltham, MA, USA) as per manufacturer’s protocol. Following RNA extraction, Qiagen’s RNeasy Plant Minikit and RNase-Free DNase Set was used for DNAse treatment following the “RNA Cleanup” protocol available in Qiagen’s RNeasy Mini Handbook (Qiagen, Toronto, ON, Canada). As sample purity was often compromised as a result of the DNAse treatment procedure, samples then underwent a sodium acetate precipitation. This precipitation used 3M C₂H₃NaO₂ (pH 5.2) and subsequent ethanol washes (100% followed by 75%) before resuspension in molecular grade water to yield RNA of increased purity.

cDNA libraries were constructed by Genome Québec following their polyA Enriched RNA Library Preparation protocol. Paired-end 100 bp reads were sequenced for a minimum of 25 million reads per library on the Illumina NovaSeq sequencing system at Genome Québec (Montréal, Québec, Canada). All sequencing data can be found at the Gene Expression Omnibus, under accession GSE284432.

Expression of PATHOGENESIS RELATED PROTEIN 1 (PR-1; LOC115704466), PATHOGENESIS-RELATED PROTEIN STH-2 (PR-10a; LOC115722015), THAUMATIN-LIKE PROTEIN 1B (PR-5; LOC115710654), and ENDOCHITINASE 2 (LOC115705823) were assessed using RT-qPCR. cDNA was synthesized using qScript™ cDNA SuperMix according to manufacturer’s instructions (Quantabio, Beverly, MA, USA). To quantify defense gene transcript abundance, RT-qPCR was run on PR1, PR-10a, PR-5, and ENDOCHITINASE 2 with the housekeeping genes TIP41-LIKE PROTEIN (TIP41; LOC115703022) and ADENINE PHOSPHO-RIOSYLTRANSFERASE 1 (APT1; LOC115713640) used as internal controls⁷⁰. Primer sequence information is found in Table S1.

We also used qPCR to determine the relative fungal load between samples. Genomic DNA was extracted from ground tissue using a modified cetyltrimethylammonium bromide (CTAB) method⁷¹. As a target for S. sclerotiorum, we used 18 S rDNA as described in Wytinck et al. (2022). Primer sequences and target loci for both RT-qPCR and fungal load qPCR can be found in Supplementary Table S1. SsoFast EvaGreen Supermix was used as per manufacturer’s instructions for both RT-qPCR and qPCR (Bio-Rad Laboratories, Hercules, CA, USA).

Raw reads were processed using computing clusters available through Compute Canada and the Digital Research Alliance of Canada (https://www.alliancecan.ca/en). Prior to read alignment, sequence read quality was first assessed using FastQC (v0.12.1; https://www.bioinformatics.babraham.ac.uk/projects/fastqc/)⁷². Paired-end read alignment was carried out using the C. sativa cultivar ‘Pink Pepper’ reference genome (NCBI RefSeq assembly GCF_029168945.1) and the S. sclerotiorum reference genome^73,74 (NCBI RefSeq assembly GCF_000146945.2; ) using HISAT2 (v2.2.1; https://daehwankimlab.github.io/hisat2/)⁷⁵. Transcript abundance was determined using featureCounts (v2.0.3; https://subread.sourceforge.net/)⁷⁶. With one of the barriers to working with the C. sativa transcriptome being the level of genome annotation, we used predicted protein orthologs publicly available for the loci of the C. sativa cultivar ‘Pink Pepper’ genome through the NCBI Genomes database (RefSeq accession GCF_029168945.1; Lim, 2023). Differential gene expression analysis, low counts filtering, library normalization, principle component analysis and further data visualization was done using libraries DESeq2 (v1.42.1; https://bioconductor.org/packages/release/bioc/html/DESeq2.html)⁷⁷, ashr (v2.2.63; https://cran.r-project.org/web/packages/ashr/index.html)⁷⁸, and ggplot2 (v3.5.2; https://cran.r-project.org/web/packages/ggplot2/index.html)⁷⁹ in R (v4.3.1; https://www.r-project.org/)⁸⁰. Genes with counts lower than 10 across all samples were filtered prior to normalization and differential gene expression analysis. Raw sequenced read counts were normalized using the median of ratios method in DESeq2⁸¹. Differentially expressed genes were called with a p-value < 0.01 when adjusted for false discovery rate (FDR) by the Benjamini-Hochberg method⁸². GO term enrichment was carried out on differentially expressed gene sets using SeqEnrich (v2.0)^83,84. GO and DEG lists (Supplementary Material 1), raw counts aligned to C. sativa (Supplementary Material 3), and raw counts aligned to S. sclerotiorum (Supplementary Material 4) are provided as supplemental datasets.

Sample preparation, sectioning and staining followed the methods previously described by Chan and Belmonte (2013) with slight modifications. Harvested colas were trimmed to the above-mentioned region of interest before being fixed in a solution of 2.5% glutaraldehyde and 1.6% paraformaldehyde in 1x phosphate-buffered saline. Tissue was added to fixative solution before being vacuum infiltrated for 30 min to ensure adequate penetration of the fixative into the C. sativa tissues. Tissue samples were fixed for 24 h at 4 °C. Tissue was decoloured in methyl cellosolve for 24 h, followed by daily 100% ethanol changes for three days at 4 °C. Historesin (Leica Microsystems, Wetzlar, Germany) was gradually infiltrated into processed tissue using a 30%, 50%, 75% and 100% ethanol: historesin mixture. Pure Historesin was exchanged three times over the course of a week, while vacuum infiltrating the tissue in for 30 min halfway through this period. Tissue was then embedded in round molds using an embedding medium composed of 91.5% Historesin, 2.4% polyethylene glycol 400, and 6.1% Historesin Hardener (Leica Microsystems, Wetzlar, Germany; Chan & Belmonte, 2013).

Hardened Historesin blocks were sectioned at 3 μm using disposable Epredia Edge-Rite steel blades (Epredia, Kalamazoo, MI, USA) mounted on a Leica RM2245 microtome (Leica Microsystems, Wetzlar, Germany). Sections were placed on glass slides for staining.

Sections were first stained with periodic acid-Schiff stain (15 min in 0.1% periodic acid, followed by 15 min in Schiff’s reagent) before being stained with 0.1% toluidine blue O suspended in distilled water for 30 s. Following staining, coverslips were mounted on the slides using Cytoseal 60 (Richard-Allen Scientific, Kalamazoo, MI, USA). Slides were viewed using a brightfield light microscope and micrographs were taken using the Leica Application Suite software version 4.6.0 (Leica Microsystems, Wetzlar, Germany). Image cropping and the addition of scale bars was carried out in Adobe Photoshop version 25.7.0 (Adobe Systems Inc., San Jose, CA, USA).