Cucumber (C. sativus cv. Kfir, Zeraim Gedera, Israel) seedlings were grown in a nursery tray with a potting mixture (Even-ARI, Israel) in controlled environment growth chambers at 22 ± 1°C. Seedlings were kept under a 12 h photoperiod and were fertilized and drip-irrigated twice a day (~0.1 l per irrigation) with 5:3:8 NPK fertilizer (N-120 mg/l; P-30 mg/l; and K-50 mg/l).

Pythium spinosum (PS-01 isolate, GenBank accession number MF116303) was cultured on PDA (Difco Laboratories) and incubated at 25°C for 2 days. Six agar disks (9 mm diameter) were cut from the periphery of the colony and placed in a 500 ml Erlenmeyer containing 80 g autoclaved pearl millet. The P. spinosum-colonized millet was incubated for 6 days at 22°C ± 1°C and then homogenized, adjusted to 0.25% (w/w) with Vermiculite (Agrekal, Israel). The control treatment contained a mix of 0.25% non-colonized millet seeds. CGMMV inoculation was performed as previously described (Reingold et al., 2015). Briefly, cotyledons of cucumber plants were gently rubbed with phosphate buffer (0.01 M, pH 7) containing carborundum dust (silicon carbide) and extract of CGMMV-inoculated cucumber leaves (Ah isolate, GenBank accession number KF155232). Non-inoculated plants and plants infected only with P. spinosum were similarly treated with virus-free buffer and carborundum.

Six days after sowing, 300 seedlings were transplanted into P. spinosum-inoculated vermiculite medium, and an additional 100 seedlings were transplanted into non-inoculated control medium. Each cucumber seedling was transplanted to a 100 ml pot (Kolbolagan, Israel) containing 100 g vermiculite and grown in greenhouse conditions described above. Five days after the P. spinosum inoculation, which permitted observation of the primary “damping off” of seedlings due to Pythium alone (Philosoph et al., 2018), 100 symptomless plants were taken to the core experiment, along with the 100 non-inoculated plants. Fifty plants from each group were then inoculated with CGMMV as described above, forming four different treatments: (i) plants infected with P. spinosum inoculated with CGMMV (PS+CG); (ii) plants with P. spinosum (PS); (iii) plants inoculated solely with CGMMV (CG), and (iv) non-inoculated healthy control plants (C).

The first collection of samples began on the same day as the CGMMV inoculation (T0); samples were taken from the collar-region (where the hypocotyl meets the root) of 5 cucumber plants inoculated with P. spinosum and 5 non-inoculated control plants to characterize the plant response before the CGMMV inoculation. Subsequently, the samples were taken from the collar-region of all the four treatments described above (5 plants per treatment) 1, 2, 3, 6 and 13 days post-viral inoculation (dpvi) with CGMMV. All samples were collected in the controlled-environment chamber directly into a test tube (Eppendorf Tubes^®, Hamburg, Germany) with liquid nitrogen and stored at -80°C for later RNA extraction. In addition, the roots of all plants were tested for the presence of P. spinosum using selective corn meal agar, as described by Philosoph et al. (2018).

At the end of the experiment, total RNA was extracted from all cucumbers’ collar-region samples (~200 mg) using the GenElute mammalian total RNA miniprep kit (Sigma-Aldrich, USA). Firstly, samples were placed in lysis buffer and mercaptoethanol with two 5-mm-diameter tungsten balls, and tissue was grounded using FastPrep-24 5G Instrument (MP Biomedicals, Santa Ana, California) at 6 m/s for 40 s for two cycles. The extraction proceeded according to the manufacturer’s protocol. Extracted RNA was treated with DNase (TURBO DNA-free Kit, Ambion Life Technologies, USA). RNA yield and purity were measured by Nanodrop (ND-1000 Spectrophotometer, Wilmington, USA) and validated for quality by running an aliquot on a Bioanalyzer 2200 TapeStation (Agilent Technologies, California). For each treatment, at each time point, libraries were prepared from up to four biological replicates. Single-end RNA-seq libraries (TruSeq RNA Library Prep Kit v2) (50 bp) were prepared and sequenced in the Technion Institute of Technology on an Illumina HiSeq 2500 machine. RNAseq raw data reads are publicly available in the NCBI under BioProject ID PRJNA808669. Raw reads (FastQ files) were inspected for quality with FastQC v0.11.5 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and trimmed for quality and adaptor removal using Trim Galore default settings. (https://github.com/FelixKrueger/TrimGalore). The average raw reads library size per sample was ~26.9 M. An average of 1.28% of the reads were trimmed. Trimmed reads were mapped to Cucumber_ChineseLong_v2 genome as downloaded from http://cucurbitgenomics.org/ftp/genome/cucumber/Chinese_long/v2/by using STAR mapper v. 2.6.0c (Dobin et al., 2013). RSEM package (Li and Dewey, 2011) was used for quantifying of genes and isoform abundances. The average percent of alignable reads was 68% from which, an average of 97% were aligned uniquely. Expected counts of mapped reads both to genes and transcripts of all samples were pooled and subjected to differential expression analysis using the DESeq2 R package (Love et al., 2014). Principal Component Analysis (PCA) were generated using R precomp function with several different subsets of samples to reduce the experiment design’s complexity.

The differential expression analysis of genes was determined based on the read counts of expressed genes using DESeq2 package in R with a significance of False Discovery Rate (FDR ≤ 0.05) (Love et al., 2014). The threshold of log2 fold change greater then > 2 or smaller than < -2, and P-adjusted (FDR) values below 0.05 was used to identify Differentially Expressed Genes (DEGs) for each treatment comparison.

To characterize the DEGs of plants infected with P. spinosum just before the inoculation with CGMMV, a comparison between P. spinosum and control at T0 was performed. For Differentially expressed genes annotation we used blastx within the Galaxy platform (https://usegalaxy.org) (Cock et al., 2015). Then we did functional classification and prediction by using Clusters of Orthologous Groups (COG) analysis (Tatusov et al., 2003) and eggnog-mapper (Huerta-Cepas et al., 2017), based on eggNOG v4.5 orthology data (Huerta-Cepas et al., 2016). Gene Ontology (GO) enrichment analysis for the DEGs was performed by AmiGO 2 via Cucurbit Genomics Database (CuGenDB-http://cucurbitgenomics.org/) with a threshold of FDR <0.05. The REVIGO program (http://revigo.irb.hr/) was used for visualization of enriched GO biological process terms (Supek et al., 2011) and the interaction between GO biological process was visualized by using Cytoscape V3.8.0 (Shannon et al., 2003). Pathways enrichment analysis of upregulated and downregulated genes was performed using Kyoto Encyclopedia of Genes and Genomes (KEGG) mapper (http://www.genome.jp/kegg/mapper.html) and CuGenDB (FDR<0.05).

To study the influence of CGMMV inoculation, a comparison was conducted between plants infected with CGMMV versus (vs) healthy control plants at each time point, starting from 24 hours after inoculation (D1) and in subsequent days (2, 3, 6, and 13 dpvi). The DEGs obtained from these comparison pathways were enriched using KEGG mapper and CuGenDB (FDR < 0.05).

The next step was to test the CGMMV inoculation effect on plants previously infected with P. spinosum, therefore comparing co-infected plants inoculated with P. spinosum and CGMMV (PS+CG) vs plants infected with P. spinosum. KEGG was used to predict the DEGs’ enriched pathways by using KOBAS to test the statistical enrichment of DEGs in KEGG pathways (Xie et al., 2011), and data were visualized in scatterplot by using Plotly (https://plotly.com/python). The DEGs data were adjusted through quantile normalization and then standardized using EXpression Analyzer and DisplayER (EXPANDER) v7.0 (Shamir et al., 2005). After normalization, the average signal value of the biological replicates for each sample was used to perform hierarchical clustering analysis and Heatmap using the ClustVis tool (https://biit.cs.ut.ee/clustvis/) (Metsalu and Vilo, 2015). Pathways enrichment analysis of upregulated and downregulated genes was performed using KEGG mapper (Mao et al., 2005) and CuGenDB (FDR<0.05). The overlapping downregulated DEGs in each pathway from “PS+CG vs control”, and upregulated DEGs in each pathway from “P. spinosum vs control” analysis, were determined by Venny 2.1 (https://bioinfogp.cnb.csic.es/tools/venny/index.html).

Total RNA was reverse‐transcribed using Verso cDNA Synthesis Kit (Thermo Scientific™, USA). Relative quantification of selected mRNA defense-related genes was performed (Livak and Schmittgen, 2001). Primers for all tested genes were designed using Primer3plus (http://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi/) and verified for their specificity with Primer Blast (http://www.ncbi.nlm.nih.gov/tools/primer-blast/). In addition, two cucumber reference genes (18S and F-box) were selected from Shoresh et al. (2005) and Migocka and Papierniak (2011), respectively. All primer sequences appear in Supplementary Table S1 . Each PCR amplification was performed for three independent biological repeats, with two technical repeats and carried out in a StepOnePlus™ RT‐qPCR (Applied Biosystems, USA), following the SYBR Green method (Fast SYBRTM Green Master Mix, Applied Biosystems) as described by Perrone et al. (2012). The PCR program consisted of an initial denaturation at 95°C for 10 min, followed by 40 cycles of 10 s at 95°C, 15 s at 60°C, and 20 s at 72°C. Gene expression data were calculated as expression ratios (quantity relative to that of control). The genes’ expression levels obtained by qRT-PCR, and those of the RNA-seq were correlated with linear regression.

To further quantify the CGMMV titer and P. spinosum concentration in the cucumber collar-region, cDNA from the samples of each treatment and time point were amplified by using CGMMV coat (Reingold et al., 2015) and movement protein gene primers, and P. spinosum specific primers for Ubiquitin and Actin ( Table S1 ). All procedures were conducted as described above. For all primer sets, efficiency was determined by a standard curve.

Samples from the cucumber collar-region of CGMMV-infected plants (with or without P. spinosum), and from healthy control plants, collected on 13 dpvi, were sliced and fixed with 4% formaldehyde and 0.2% glutaraldehyde, as described previously (Reingold et al., 2015). The slices were washed twice with PBS-T (phosphate-buffered saline with 0.05% Tween-20), blocked using PBS with 1% milk (0% fat) for 30 min and incubated overnight at 4°C with IgG antibodies specific for CGMMV (Antignus et al., 1990; Antignus et al., 2001). Slices were washed twice with PBS-T, and the secondary antibody, goat anti-rabbit IgG-conjugated Alexa Fluor 488 (Invitrogen), was added to the slices in a 1:1,000 dilution in PBS followed by incubation at 37°C for 3 h. The slices were washed twice with PBS-T and kept in PBS in a sealed box. The fluorescence signal of the slices was detected using confocal microscopy (LSM510 Axiovert 100 M, ZEISS, USA).

FISH was conducted according to Gottlieb et al. (2006) with several modifications. Single-stranded DNA oligos of 20 nucleotides (5-GAACCAGTACGACCCTCCAA-3) were labeled with a Cy3 fluorophore and used as a probe. The probe (Hylabs, Israel) corresponded to the P. spinosum Actin gene. Collar-region samples were collected from P. spinosum-infected plants with or without CGMMV co-infection and from control plants, and were hand-sectioned, followed by fixation overnight at room temperature in Carnoy’s fixation (6:3:1 v/v/v chloroform: absolute ethanol: glacial acetic acid). Samples were decolorized twice in 6% H₂O₂ solution in absolute ethanol for 48 h, then washed twice in absolute ethanol and pre-hybridized with hybridization buffer (20 mM Tris–HCl, pH 8.0, 0.9% NaCl, 0.01% sodium dodecyl sulfate, 30% formamide) for 1 h at room temperature. Ten pmol fluorescent probe ml^-1 was added to the hybridized samples followed by overnight incubation at room temperature. Samples were then examined using a confocal microscope (LSM510 Axiovert 100 M; ZEISS, USA).

Five days after the P. spinosum inoculation (which simulated the process of early damping-off stage with Pythium alone), 188 wilting cucumber plants were discarded; 100 inoculated plants and an additional 100 non-inoculated healthy control plants served the core experiments. Fifty cucumber plants from each group were then inoculated with CGMMV as described above, forming the four different treatments described above: (i) PS+CG; (ii) PS; (iii) CG, and (iv) C. The inoculation of the plants with CGMMV is considered as the start of the experiment (T0) and only minor growth constraints were observed in the Pythium-established plants ( Figure 1A ). No plant mortality was detected in any treatment until day 6 dpvi. The first wilting symptoms were observed only in the combined infection treatment (PS+CG) ( Figure 1B ), reaching 33.3% mortality between 6-13 dpvi (12 out of 36 plants) and the surviving co-infected plants suffered from severe growth inhibition ( Table 1 , Figures 1B, C ). The first viral symptoms of mottle and mosaic appeared on day 10 and were well apparent on day 13 without any visual differences between the CG and PS+CG treatments. No wilting symptoms were observed in the PS, CG, and C treatments ( Table 1 , Figures 1B, C ).

To gain insight into the genes involved during the infection process of P. spinosum followed by CGMMV infection in cucumber plants, samples of collar-region from the four treatments (P, P+CG, CG, C) were collected in six-time points: pre (T0) and post-CGMMV inoculation (1, 2, 3, 6, and 13 days). Data obtained from the transcriptomic analysis presented an average of 26,936,546 reads per sample ( Table S2 ). High quality reads (97%) were mapped to C. sativus CoDing Sequences (CDS) reference. Overall, 20,792 different transcripts were identified ( Table S3 ). A principal component analysis (PCA) of all collar-region transcriptional changes divided the samples into two main clusters. The first cluster includes 23 samples from all the healthy control plants and also from the plants infected only with CGMMV; the second cluster includes all 30 samples infected with P. spinosum with and without CGMMV, regardless of the sampling time. Most of the variance was defined by PC1 (61%) and PC2 (8%) ( Figure 2 ).

To reveal the changes that occurred in P. spinosum-infected plants following CGMMV inoculation, plants co-infected with CG+PS vs PS alone for each time point were compared (1, 2, 3, 6 and 13 dpvi). In total, DEGs were downregulated that related to defense mechanisms against necrotrophic pathogens were detected 1 dpvi (of P. spinosum-infected plants). 263 DEGs with (P_adj ¾ 0.01, -2 ≥ log₂fold ≥ 2) were detected, among them, 209 DEGs already differentially expressed as early as 1 dpvi, of those 206 DEGs were downregulated. No major changes in DEGs were detected during days 2-6. However, on day 13, 52 DEGs were upregulated while only two DEGs were downregulated ( Figure 5A , Table S7 ).

While assigning the DEGs to the KEGG pathway analysis, among the top 25 pathways, 21 were enriched (P < 0.05) and were down-regulated on 1 dpvi) Figure 5B ). The most significant number of DEGs were assigned to ‘phenylpropanoid biosynthesis’ (n=24), ‘biosynthesis of secondary metabolites’ (n=46), ‘phenylalanine metabolism’ (n=13) and ‘linoleic acid metabolism’ (n=4) KEGG pathways ( Figure 5B ), all belonging to defense mechanisms against pathogens. In contrast, the KEGG analysis of the upregulated DEGs from day 13 were only enriched for ‘lysine biosynthase’ pathway (n=1) ( Figure 5C ). However, additional DEGs were assigned to the non-enriched pathways of ‘plant-pathogen interactions’ (n=1), ‘phenylpropanoid biosynthesis’ (n=1) and ‘biosynthesis of secondary metabolites’ (n=2) ( Figure 5C ).

We compared the 263 specific DEGs derived from CGMMV inoculation of established-Pythium infected plants (PS+CG) for all of the 53 experimental samples. Three main patterns were detected by using the expander software ( Figure 6 , Table S7 ). Cluster 1 included 26 genes that were downregulated within all samples with Pythium (with/without CGMMV) vs. controls and CGMMV alone. There was no distinctive pattern for these genes.

Cluster 2, including 181 genes mostly not expressed in the control or CG plants. However, the cluster had a clear high expression pattern in plants infected with Pythium alone. Interestingly, 118 DEGs were strongly downregulated 1 dpvi ( Figure 6 , Table S8 ). The cluster 2 pattern was constant for most DEGs throughout the 13 days in plants infected with both CGMMV and Pythium.

Eight enriched pathways were obtained from the 181 DEGs of cluster 2. To learn the direct influence of the CGGMV penetration into the established-Pythium infected plants, the eight enriched pathways were compared to the 10 enriched pathways derived from plant infected with Pythium alone (prior to CGMMV inoculation, i.e. Figure 4A ). Six shared enriched pathways were obtained, which included mutual DEGs that were upregulated with Pythium alone, and strongly down regulated after the CGMMV inoculation ( Figure 7A ). All of these six mutual pathways are involved in plant defense mechanisms against necrotrophic pathogens including phenylpropanoid (22 mutual DEGs); jasmonic acid (16); phenylalanine (13); ethylene (11); suberin monomer biosynthase (10); scopolin and esculin biosynthesis (8). In addition, two pathways, L- phenylalanine (5 DEGs) and L-glutamate (2 DEGs) were enriched only during combined infection ( Figure 7A ). Among the 10 enriched pathways, four were solely upregulated with Pythium alone [glutathione metabolism (13 DEGs), oxalate degradation (10), pyruvate metabolism (7) and plant pathogen interaction (5)]. Although these four last pathways were not enriched during the co-infection, they also include DEGs that were downregulated at 1 dpvi ( Figure 7A ).

Relative expression of 7 selected down-regulated genes known to be related to defense pathways against necrotrophic pathogens [JA, phenylpropanoid, scopolin, ethylene and plant-pathogen interaction] were validated. The qRT-PCR results confirmed the significant down-regulation (P < 0.05) for all the seven selected genes 1 dpvi vs established-Pythium infected plants (prior to CGMMV inoculation, Figure 7B ). Furthermore the qRT-PCR results also correlated well with the RNA-Seq data analysis (R^2 = 0.807, P < 0.05, Figure S3 ).

Cluster 3, includes 42 genes upregulated only 13 dpvi. Moreover, all the 42 genes were differentially expressed in the combined infection, while 26 of them were also differentially expressed in CG plants ( Table S8 ). These genes are mostly related to defense mechanisms against viral infection including RdRp (2 DEGs), WRKY transcription factors (5 DEGs) and pyrimidine metabolism (5 DEGs). RNA-dependent RNA polymerase (RdR1c1) and two selected genes from pyrimidine metabolism:AAA-ATPase (BCS1-ATOM66) and ATP-dependent zinc metalloprotease ftsH (BCS1) were validated by qRT-PCR ( Figure S4 ). All these genes are upregulated following CGMMV infection (13 dpvi), and therefore this data accords with the RNA-Seq data analysis (R^2 = 0.986, P < 0.05). An additional minor fourth Cluster included 3 genes with an unspecified pattern ( Table S8 ).

CGMMV titer in the collar-region increased from day 6 to day 13, with or without the presence of P. spinosum, indicating that P. spinosum did not affect CGMMV titer (accumulation or decrease) in co-infected plants ( Figure 8A ). This was validated by the in situ immunofluorescence analysis on 13 dpvi, as no substantial differences were visualized between CGMMV vs PS+CG plants ( Figure 8B ). In contrary, relative gene expression ratio of the P. spinosum-Actin in the collar-region shows that in plants infected with P. spinosum alone, the Actin concentration remained constant during the next 13 days compared to the Actin concentration at T0. However, a 36-fold increase of P. spinosum-Actin was detected at 13 dpvi in plants co-infected with P. spinosum and CGMMV when compared to plants inoculated with P. spinosum alone ( Figure 8C ); similar results were also detected for the P. spinosum-Ubiquitin ( Figure S5 ). FISH analysis of the P. spinosum-Actin gene also visually demonstrated the increase of the P. spinosum hyphae in the PS+CG plants when comparing to PS infected plants (collar region) on 13 dpvi ( Figure 8D ).

The results support our hypothesis and confirm the scenario of a unique situation in planta that provided optimal conditions for re-proliferation of P. spinosum during co-infection.

In contrast to the late reaction in the plant collar-region in response to CGMMV inoculation into naive plants, CGMMV inoculation into established-Pythium plants elicited a rapid host response as early as 1 dpvi. Moreover, insight into the specific DEGs of this co-infection scenario revealed a sharp downregulation in several pathways related to plant immune responses to the necrotrophic pathogen. Moreover, while comparing those DEGs to those upregulated in the Pythium-established plants (on T0), not only those DEGs involve the same pathways, but also included a large number of mutual genes ( Figure 7 ). Some of these pathways such as JA/ET are known to take part in cross-talk with SA (Koornneef and Pieterse, 2008).

CGMMV is a biotrophic obligate parasite, that stimulates plant responses through the SA pathway (Yang et al., 2015), while Pythium is a necrotrophic pathogen. Previous studies show that simultaneous activation of both SA and JA/ET pathways in the same host is mostly exhibited by suppressing the JA/ET pathway (Spoel et al., 2007; Pieterse et al., 2012). Although we do not observe any rapid response in the plant collar-region to the biotrophic pathogen inoculated in the leaf (e.g., SA pathway-related genes), a rapid suppression of pathways involved in the response against necrotrophic pathogens was apparent. This may suggest that the cross communications occur and result from systemic reactions, as previously described by Caarls et al. (2015). Their review described potential regulatory mechanisms that suppress the JA pathway in the presence of SA via several transcription factor genes, such as specific members of the bHLH and WRKY family (Caarls et al., 2015; Aerts et al., 2021).

While following the pattern of the 263 DEGs obtained from the co-infected plants, throughout all the treatments in the experiment period, three main dynamic patterns are observed that were corroborated with each phenotypic characterization of the plant responses. Genes in Pythium-established plants kept relatively high expression level until 6 dpvi (11 days after Pythium inoculation), but with fluctuation between time points ( Figure 6 ), a phenomenon previously observed in another Pythium-infected pathosystem (Shin et al., 2014; Shin et al., 2016). From 6 dpvi onward, those DEGs showed a milder expression pattern. This molecular pattern is in line with the phenotypic aspects, as these Pythium-established plants only showed moderate growth constraints and very limited mortality incidence in our experiments and in previous studies (Philosoph et al., 2018).

The co-infected plants demonstrated two other patterns. The second pattern involved a group of DEGs related to the defense mechanisms against necrotrophs and demonstrated a sharp downregulation as early as 24 hours post-CGMMV inoculation. Most of these genes maintained a relatively low expression throughout the experiment, although several genes were suddenly upregulated 13 dpvi. The phenotype of the co-infected plants supported the gene response pattern, demonstrating severe stress symptoms, including 30% mortality, while the rest of the plants suffered from extreme growth constraints several days before the appearance of the viral symptoms ( Figure 1B ). The third pattern involved genes that were upregulated only 13 dpvi. Most of these genes were expressed with the same pattern as in plants infected with CGMMV alone. This pattern was well correlated with a significant amount of viral titer quantified in the collar-region ( Figure 8A ), along with mottle and mosaic symptoms that appeared 13 dpvi ( Figure 1C ). These data are supported by in situ hybridization ( Figure 8B ), all the results supporting the claim that Pythium does not influence directly CGMMV accumulation.

Quantification of Pythium in the plants reveals that the pathogen’s concentration significantly increased on 13 dpvi vs PS ( Figure 8C ) and supported by the FISH results ( Figures 8D, E ). These results may explain the plant collapse after the co-infection process. Moreover, several upregulated DEGs 13 dpvi by the co-infected plants are related to defense mechanisms against necrotrophic pathogens such as ACC, LOX and phospholipase a1 known to be stimulated by the JA/ET pathway (Van Loon et al., 2006; Wasternack and Song, 2017). These signs of recovery in the plant defense may imply a renewed battle of the host against Pythium, after the destabilization following CGMMV infection. It also opens a new research question about the boundaries of the time windows examined and the ability to depict plant susceptibility. The importance of such a time window can significantly contribute to optimizing pest management against co-infection. For example optimizing anti-pythium compounds application timing to just before the agricultural practices that spread the CGMMV (triseling or leaf cutting), or application of compounds that increases plant defense against necrotrophic pathogens (Dempsey et al., 2022; Yu et al., 2022), during the most sensitive stages.