Potential mechanisms of quantitative resistance to Leptosphaeria maculans (blackleg) on cotyledons of canola ( Brassica napus ) CURRENT STATUS: POSTED

Background: Blackleg disease, caused by Leptosphaeria maculans ( Lm ), can lead to significant losses of canola/rapeseed crops. Growing resistant canola cultivars can be an effective and environmentally friendly way to manage blackleg. Major resistance genes may stop infection, but can also be rapidly overcome by shifts in pathogen population towards virulence. Thus, using race-nonspecific or quantitative resistance (QR) is of interest because it is potentially more durable. However, the mechanisms and genes underlying QR are mostly unknown. In this study, we explored QR in “74‑44 BL”, a Canadian canola cultivar carrying a moderate level of race nonspecific resistance, based on cotyledon inoculation (Supple. Fig.1) . The susceptible cultivar “Westar” was used as a control. Lesions developed more slowly on 74-44BL than on Westar. We used RNA-Seq to identify genes and gene functions putatively involved in the QR. Results: Relative to inoculated Westar, some of the B. napus genes that were differentially expressed strongly in inoculated 74-44 BL included those putatively involved in programmed cell death (PCD), reactive oxygen species (ROS) generation, signal transduction and/or intracellular endomembrane transport. Examples included genes annotated as a Bax inhibitor 1, a development/cell death (DCD) domain containing proteinases and peptidases, all of which could play a role in PCD and a zinc-finger Sec23/Sec24 and five small GTPases likely involved in endoplasmic reticulum (ER) to Golgi vesicle traffic and/or signal transduction. Further experiments, however, did not confirm changes in genomic DNA degradation, a potential marker for PCD, between the two cultivars. In addition, infection progression in cotyledons was not altered by applying protease inhibitors directly to cotyledons. Additional testing was done using green fluorescent protein (GFP)-tagged Lm for cotyledon colonization as well as ROS production, in relation to the lesion development. The results showed that ROS production occurred beyond the area colonized by Lm hyphae in 74-44 BL. Conclusions: ROS may also be involved in signal transduction and/or intracellular endomembrane transport. These results provide a starting point for a better understanding of the mechanisms behind QR against Lm in canola and developing new host-resistance strategies for management of blackleg.

Fluorescent microscopy of proteins tagged with fluorophores, such as green fluorescent protein (GFP), provides valuable information about plant colonization by microbes, including the canola-blackleg pathosystem [12,20]. Next generation sequencing approaches may help relate phenotypic observations, such as those obtained from microscopy, to molecular mechanisms. Here we present data on the colonization and lesion formation in Westar (susceptible) and  cotyledons inoculated with a GFP-expressing isolate of L. maculans. This work also aimed to explore the genes differentially expressed at the seedling stage between canola cultivars in order to gain insights into the potential mechanisms of QR in 74-44 BL.

Methods
This manuscript includes the following experiments on cotyledons of canola cultivars without (Westar) and with QR (74-44 BL): 1) RNA-Seq and corresponding infection severity, 2) time series evaluation of lesion size and the corresponding area colonized by Lm hyphae, 3) staining for the reactive oxygen species (ROS), hydrogen peroxide, via 3,3-diaminobenzidine (DAB), 4) a protease inhibitor study and 5) an assessment of the level of fragmentation of genomic DNA as a proxy for programmed cell death (PCD).

Fungal and plant material
Inoculum was prepared from L. maculans isolates 12CC09 carrying AvrLm6,7 and 12CC09-GFP, grown on V8 agar until pycnidia were visible. Isolate 12CC09-GFP was generated by transforming the isolate 12CC09 with a binary vector containing the GFP gene via Agrobacterium-mediated transformation.
Pycnidiospores were harvested in sterile water, filtered through a Falcon™ Cell Strainer (70 μm pore size), diluted to 2 ×10 7 spores / mL and stored at -20°C until use. One week after planting, cotyledons were wounded on each lobe with modified tweezers before being inoculated with 10μl droplets of water or pycnidiospore suspension. 74-44 BL is DEKALB ® hybrid with multi-genic Lm resistance and R genes Rlm1, Rlm3 and RlmS (Saskatchewan Seed Guide, 2019). This cultivar also carries a level of QR against multiple Lm races in cotyledons [11,12] found that 74-44 BL carried both race nonspecific resistance and specific R genes Rlm1, Rlm3 and Rlm9. The QR was expressed in cotyledons in terms of both lower lesion scores (see Table 1 of Hubbard and Peng [11]) and more restricted Lm colonization [12]. Plants were grown in Sunshine #3 soil-less mix (Sun Gro Horticulture Canada Ltd., Vancouver, BC) to which 12.5 g L -1 Osmocote Plus 16-9-12 (N-P-K; Scotts Miracle-Gro Canada, Mississauga, ON) had been added. For all experiments, except those involving the time series that did not involve DAB staining to detect ROS, canola plants were grown in 72-well flats and placed in a growth chamber set to 22°C and 16°C during the 16 hours of light (approximately 280-575 μmol m -2 s -1 ) and 8 hours of darkness, respectively. Plants intended for the time series microscopic examination were grown either as described above or in the greenhouse in 10 cm square pots, exposed to a mix of natural and fluorescent (430W Philips high pressure sodium lamps) light, and inoculated with water, 12CC09 or 12CC09-GFP. Isolate 12CC09 was included as a control to determine if fluorescence observed could be attributed to GFP.
Plants were divided into Lm-inoculated and mock-inoculated. Within each inoculation treatment, plants were split between cultivars. The RNA-Seq and time-series microscopy experiments were repeated three times, as were the experiments that involved staining for hydrogen peroxide (ROS) with DAB. The protease inhibitor experiments were carried out five times.
For RNA-seq experiments, within each replicate, there were six seedlings per treatment , divided at random into two blocks of three plants. At 7 days post inoculation (dpi), cotyledon samples were taken for RNA extraction and subsequent RNAseq, from three of these seedlings. The other three seedlings were maintained until 14 dpi and rated for infection severity on the 0-9 scale [21,22].

RNA extraction, library preparation and sequencing
Samples, measuring 5-10mm × 5-10mm, were collected from the area adjacent to and containing the lesion on each lobe of the cotyledons at 7 dpi (Fig. 1A). Samples were flash frozen in liquid nitrogen and stored at -80°C to await RNA extraction. Samples from one lobe (lobe 1, 2 or 3) were pooled from three replicates, each containing three seedlings, for a total of nine samples per RNA extraction. RNA was only extracted from one of the inoculated lobes.
Cotyledon tissue was ground in liquid nitrogen by vortexing in 50mL Nalgene Oak Ridge tubes containing two metal balls. RNA was extracted from 40-50 mg of the ground and frozen tissue using the QIAGEN RNeasy Plant mini kit on a QIAcube with a DNase I on-column digestion. The concentration and integrity of the resulting RNA was assessed via Nanodrop and Experion (Bio-Rad using the DESeq2 package [26]. Genes were considered differentially expressed if they had a log base 2 (log 2 ) fold change in expression above 2 or below -2 and an adjusted p-value under 0.05.
Enrichment analysis based on gene ontology (GO) terms was performed by using the Blast2Go-pro suite [27]. All B. napus genes were searched against the non-redundant protein database from National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/) using BLASTX algorithm with an E-value threshold of 1e -5 . All BLAST hits were then mapped to the GO database to retrieve GO terms that associated with each hit. Subsequently, all B. napus genes were searched against the InterPro database (http://www.ebi.ac.uk/interpro/) and annotated by merging the Blast2GO and InterPro results. GO terms that were significantly enriched in DEGs were identified by comparing with the whole genome background with a false discovery rate (FDR) ≤ 0.05.

Time series infection in cotyledons examined with microscopy
Both fluorescent and bright-field images were collected with a Zeiss Stereo-Lumar epifluorescence microscope, equipped with a NeoLumar S 0.8× objective and an Axiocam 512 camera. Light was provided by a KL-2500 LCD bulb and a HBO100 mercury lamp for bright field and fluorescent microscopy, respectively.
Two separate sets of time series microscopy experiments were carried out. In the first set of experiments, cotyledons were detached from the plant for imaging at 3, 7, 10 and 14 dpi. In the second set of experiments, imaging was done at 3, 5, 7, 9 and 11 dpi. The second set of experiments also included colorimetric staining for hydrogen peroxide. For both sets of experiments, different plants were used at each time point. Because the lesion area and the area colonized by Lm hyphae at 3 dpi were frequently zero, or close to zero, data from this time point is not presented.
For each inoculation site, parameters were measured with the aid of ZEN 2 pro and/or ZEN 2.3 lite (blue edition, © Carl Zeiss Microscopy GmbH, 2011) software. The software automatically takes the image magnification into account. Bright field images of the top surface of each cotyledon were used to measure the area of the lesion (mm 2 ), the distance from the inoculation point to the most distant edge of the lesion (mm) and the area stained hydrogen peroxide (mm 2 ). The area colonized by hyphae (mm 2 ) and the distance from the edge of inoculation wound to the furthest hyphal tip (mm) (first set of time series experiments only) were quantified using fluorescent images. The Zen Active Contour and/or Polygon Contour and Length tools were used to collect area and distance data, respectively.

Colorimetric detection of hydrogen peroxide
The area staining for the reactive oxygen species (ROS) hydrogen peroxide in B. napus cotyledons infected by Lm isolate 12CC09-GFP, was measured at 7 dpi. Lesion area and area colonized by GFPtagged Lm hyphae was measured as described above for the time series infection experiment.
Subsequently, detached cotyledons were placed in a solution of DAB at room temperature. After 40 (first two experiments) or 90 (third experiment) min, the cotyledons were vacuum infiltrated with DAB for approximately 2 to 3 hours and then boiled in 95% ethanol for approximately 10 to 20 min at 70°C to remove the chlorophyll, making the DAB staining more visible. The cotyledons were stored in 95% ethanol prior to measurement of the area stained brown for hydrogen peroxide under a dissecting microscope.
Assessment of genomic DNA degradation as a marker of programmed cell death Samples of canola cotyledons were collected as described for RNA extraction. The samples were freeze dried, and ground to a fine powder in 2mL tubes with one 3 mm tungsten carbide bead per tube in a TissueLyser (Qiagen) for 5 min at 25 hertz at room temperature. Genomic DNA was extracted using the QIAGEN DNeasy Plant mini kit according to the manufacturer's instructions.
Extracted DNA was diluted to 50 ng/μl. The integrity of the resulting DNA was assessed using Experion DNA 12K analysis kit (Bio-Rad Canada) on an Experion automated electrophoresis system according to the manufacturer's instructions.

Statistical analysis
Statistical analyses were done using SAS (version 9.3). Data were assessed for homogeneity of variance and normality, respectively, using Bartlett's Test and Shapiro-Wilk Test. Data from mockinoculated plants, which consisted exclusively of zeros, were excluded from statistical analysis.
For ratings of infection severity in parallel with RNA-seq, when only the inoculated plants are considered, a randomized complete block design (RCBD) was used, with a total of 9 replicates (3 per experiment) and 3 subsamples (plants) per replicate (Fig. 1B). This data was pooled from all plants in a given experiment and y=log 10 (x+10) transformed. Means were compared via a t-test using proc GLM. For the time series experiments based on microscopic examinations that did not include colorimetric hydrogen peroxide detection (Fig. 1E), data was subjected to a y=log 10 (x+10) transformation prior to statistical analysis. Each parameter (lesion area, area colonized by GFP-tagged Lm hyphae, distance to the edge of lesion and distance to furthest hyphal tips from the inoculation site) and time point (7, 10 or 14 dpi) were analyzed individually, using proc GLM to compare means between the two cultivars with a t-test. Because much of the data on lesion area, area colonized by Lm hyphae and area of hydrogen peroxide detection at 7 dpi ( Fig. 5A) and on the same three parameters for the microscopy time series observations (Fig. 6) did not meet all of the assumptions of ANOVA, the data was analyzed by means of the Wilcoxon two-sample (for cultivar) or Kruskal-Wallis multi-sample test (for comparisons between parameters) using proc NPAR1WAY. When significant differences were found for a given factor (cultivar or parameter), differences between individual means for combinations of treatments were assessed, using the ranked data. This was done using Tukey's multiple comparison adjustment, using lobes of a given plant as sub-samples, via proc GLM.
For the protease inhibitor experiment (Fig. 8), a two-factor analysis was performed, with cultivar (Westar and 74-44 BL) and protease inhibitor as the two factors. Data were log 10 (x+10) transformed before analysis. Lobes were considered sub-samples. Tukey's adjustment was used for multiple comparisons in proc GLM.

Infection symptoms and Lm hyphal growth in cotyledons
Among the seedlings inoculated and grown in parallel with those used for RNA-Seq, Westar showed higher infection ratings than 74-44 BL at 14 dpi (Fig. 1B). In separate experiments, however, the appearance and size of lesions, as well as the distance from the inoculation wound to lesion edge, were similar between the two cultivars at 7 dpi, while the area colonized by Lm hyphae and the distance from the inoculation site to the most distal hyphal tips were greater in Westar ( Fig. 1C and E). By 10 and 14 dpi, all of the measurements had become greater in Westar than 74-44 BL ( Fig. 1D and E).

Gene expression in L. maculans
When the criteria of adjusted p-value ≤ 0.05 and log 2 fold change ≥ 2 in expression were applied, there were only 16 differentially expressed Lm genes between inoculated Westar and 74-44 BL. Three DEGs were more highly expressed in inoculated 74-44 BL as compared to inoculated Westar (Table 1) while thirteen DEGs showed the reverse trend ( Table 2).
The Lm DEGs were all expressed at low levels. When all Lm genes were considered, there were eight genes with basemean expression values over 10,000 (ranging from 14,346 to 40,534). Three genes had expression values between 1,000 and 9,999, 85 between 100 and 999 and 152 between 50 and 99. There were a total of 12,119 genes with non-zero expression values. The most highly expressed DEG had a basemean of 40.
The three DEGs upregulated in Lm inoculated 74-44 BL had sequence similarities to genes encoding a short chain dehydrogenase/reductase, a pyoverdine biosynthesis and a hypothetical protein, respectively. Pyoverdine is a siderophore biosynthesized by Pseudomonads [28]. Zwiers et al. [29] found a gene encoding an ABC-transporter with a pyoverdine biosynthesis motif in the fungus

Genes upregulated in inoculated 74-44 BL
There were 908 DEGs upregulated in inoculated 74-44 BL, relative to inoculated Westar, but not differentially expressed between any other pairs of treatments. Two DEGs showed basemean expression levels over 10,000, six had basemeans between 5000 and 9999, and 65 had basemeans between 4,999 and 1,000.
Five DEGs with similarities to peptidases were among those with the highest scores. Indeed, the three DEGs with the highest scores were all putative peptidases. The DEG with the highest score is BnaA01g17570D, which has InterPro domains suggesting it is a cysteine peptidase belonging to family C1, sub-family C1A, papain family. Another DEG with a high score is BnaA09g52180D, a putative cysteine peptidase. The legumain peptidase C13 (BnaA01g04000D), also known as a vacuole processing enzyme (VPE), and BnaC02g00130D which has similarity to a protease involved in the degradation of Rubisco, were also upregulated. Additionally, numerous chlorophyll A-B binding proteins showed very high basemeans and were more highly expressed in inoculated 74-44 BL than inoculated Westar.
An ATPase of AAA-type, with protein BLAST similarity to RuBisCO activase, was also differentially expressed. The protein BLAST results also indicate that this DEG is potentially involved in endoplasmic reticulum (ER) to Golgi membrane budding.
Glycoside hydrolases, including a beta-galactosidase (BnaA04g04110D) and an alpha-1,6glucosidases, pullulanase-type (BnaA10g25820D) are differentially expressed, with very small adjusted p-values. A putative lactate/malate dehydrogenase (BnaC02g00740D) is also differentially expressed, albeit with a less significant adjusted p-value and higher basemean expression than BnaA04g04110D or BnaA10g25820D. The DEG BnaA03g11710D, with a thiazole biosynthetic enzyme InterPro domain, also has protein sequence similarity to a ribulose-1,5-biphoshate synthetase. Table 3 and Fig. 3 summarize the putative functions of DEGs that are more highly expressed in inoculated 74-44 BL as compared to inoculated Westar.
GO term enrichment analysis of these 908 DEGs was consistent with the results presented in Table 3 and Fig. 3 in which many of GO terms with the lowest FDR were related to photosynthesis and light responses. Furthermore, three GO terms were linked to hydrogen peroxide. While none of the enriched GO terms suggested peptidase activities, the GO term with the second lowest FDR was associated with cysteine biosynthesis (Table 5). This is consistent with the putative cysteine peptidase activity of BnaA01g17570D (Table 3).

Genes upregulated in inoculated Westar
A total of 640 DEGs were more highly expressed in inoculated Westar as compared to inoculated 74-44 BL, but not differentially expressed when any other pair of treatments were compared. The expressions of these DEGs ranged from a basemean of 3,410 to 1.25, with only 11 DEGs showing basemeans over 1,000. Twenty eight DEGs had basemeans between 500 and 999, while 73 had basemeans between 100 and 499. The remaining 527 DEGs had basemeans under 100.
The DEG with the highest score, BnaC09g20030D, showed similarity to a Bax inhibitor-1.
BnaCnng58090D, a DEG with a basemean of 2,354, is similar to a development/cell death domain (DCD). BnaC08g42820D is a DEG similar to a heat shock protein 70. BnaA04g06220D and BnaA09g26960D have similarities to Sec23/Sec24 and Sec61/SecY, respectively. Sec23 and sec24 are part of the coat protein II (COPII) complex, involved in ER to Golgi vesicle transport [32]. Five DEGs, BnaA08g26550D, BnaA06g05280D, BnaC06g24690D, BnaA07g09950D and BnaCnng06680D appeared similar to small GTPases. These DEGs have basemeans ranging from 972 to 3,100. Table 4 and Fig. 3 summarize these DEGs.
GO terms related to the ER, ER stress, vesicle transport and the cellular endomembrane system were enriched. None of the enriched GO terms, however, were associated with PCD. One enriched GO term was related to response to hydrogen peroxide (Table 6). BnaCnng58090D is not associated with any GO terms.
Hydrogen peroxide in cotyledons RNA-seq results suggested that ROS, such as hydrogen peroxide, may play a role in the QR to Lm carried by 74-44 BL. To validate this finding, DAB staining was used to quantify the area of ROS production surrounding the infection site.

Hydrogen peroxide at seven days post inoculation
The size of visible lesion, area of hyphal colonization and area with ROS detection in cotyledons varied, depending on the cultivar and parameter measured. In inoculated Westar, the area colonized by hyphae (as visualized by GFP fluorescence) and area staining positive for hydrogen peroxide were both larger than the area of necrotic lesions, while the former two parameters were not different from each other (Fig. 4A). In contrast, the lesion size and area colonized by GFP-tagged Lm hyphae did not differ in 74-44 BL, whereas the area with ROS staining was bigger than that of former two. As with the results in Fig. 1, the lesion size did not differ between Westar and 74-44 BL at 7 dpi, while the area colonized by Lm hyphae was substantially greater in Westar. The area with ROS staining did not differ between the cultivars at 7 dpi either (Fig. 4).
Hydrogen peroxide time series experiment When examined over time post inoculation, most of the parameters measured tended to increase over time. Westar and 74-44 BL responded differently to the Lm infection. In Westar, the lesions were consistently smaller than either the area colonized by Lm hyphae or the area with ROS staining (Tukey adjusted p ≤ 0.05) (Fig. 5). In 74-44 BL, however, the area stained for hydrogen peroxide was larger than that occupied by fungal hyphae or visible lesions (Tukey adjusted p ≤ 0.05), and the area of pathogen colonization was either smaller than (11 dpi) or not different (5, 7 and 9 dpi) from the size of lesion (Tukey adjusted p ≤ 0.05; Fig. 5).

Genomic DNA degradation as an indicator of programmed cell death
Because the RNA-seq results suggested that PCD could play a role in QR to Lm in 74-44 BL, we examined degradation of genomic DNA as a proxy for PCD. No difference in genomic DNA degradation was apparent between any of the treatments by either agarose gel electrophoresis or Experion 12K ( Fig. 6).

Impact of protease inhibitors on Lm infection of cotyledons
Results from the RNA-seq experiments led us to hypothesize that proteases could contribute to 74-44 BL QR to Lm. We attempted to test this hypothesis by treating cotyledons with several protease inhibitors. The direct application of protease inhibitors to surface of Westar or 74-44 BL cotyledons did not have a significant impact on either the lesion size or the area colonized by Lm hyphae within a given cultivar. However, the latter was consistently greater in Westar than in 74-44 BL cotyledons, regardless of the protease inhibitor used (Fig. 7).

Discussion
It is generally thought that QR to Lm is not expressed in canola cotyledons [2]. In this study, however, we found that the infection severity on Lm-inoculated cotyledons differed quantitatively between Westar (susceptible) and 74-44 BL (with QR). In addition to larger lesions, the area of Lm hyphal colonization was greater in Westar than in 74-44 BL; often the hyphal growth extended beyond the borders of visible lesions in Westar, while this was not the case for 74 33] also measured QR to Lm in young B. napus plants and found, in some cases, restricted Lm growth correlated with reduced blackleg in more mature plants [33]. Huang et al.
[33] also found partial overlap in quantitative trait loci (QTL) contributing to Lm resistance at both plant developmental stages. We used RNA-seq to explore DEGs between inoculated Westar and 74-44 BL as a first step to understanding the molecular mechanisms of this cotyledon-stage QR.
Many of the highest scoring DEGs, upregulated in inoculated Westar, against inoculated 74-44 BL, relate to the control of PCD, endomembrane vesicle trafficking between the ER and Golgi, as well as molecular chaperones, cation transporters, protein glycosylases and degradation enzymes ( Table 4).
The GO terms enriched in these DEGs also suggest a role in endomembrane vesicle transport ( Table   6). The gene BnaCnng58090D, with sequence similarity to a DCD domain, was upregulated in inoculated Westar (Table 4). DCD domains can stimulate a hypersensitive response, considered a form of PCD in plants [34]. Other upregulated genes with putative roles in endomembrane transport to/from the ER are potentially related to ER stress, which can trigger DCD-mediated PCD [35].
BnaC09g20030D, which is similar to a Bax inhibitor-1, was also upregulated. Bax inhibitor-1 inhibits PCD [36]. Hence, it seems reasonable to hypothesize that the cotyledon infection triggers the expression of BnaCnng58090D in susceptible plants, but that the hypersensitive response that a DCD would otherwise promote may be prevented by the activation of BnaC09g20030D, the Bax inhibitor-1.
However, the lack of differences observed in genomic DNA degradation argues against the above hypothesis. Consistently, the GO enrichment analysis did not uncover any GO terms related to PCD (Tables 5 and 6). Fragmentation of genomic DNA can be associated with plant PCD, including PCD that mimics apoptosis in animal cells, and is involved in normal plant developmental processes. When 74-44 BL was inoculated, numerous peptidases were more highly expressed than in inoculated Westar. Specifically, putative papain cysteine peptidases (BnaA01g17570D and BnaA09g52180D) likely found in the plant vacuole [42,43], as well as putative legumain peptidase C13 (BnaA01g04000D), also known as a vacuole processing enzyme (VPE), were upregulated. VPEs are, as suggested by the name, located in plant vacuoles [44,45], as shown in Fig. 8. In addition, BnaC02g00130D, which has similarity to a protease involved in the degradation of RuBisCO, is also upregulated in inoculated 74-44 BL. These genes may also be involved in PCD [reviewed by Zamyatnin [46]]. During PCD, the plant cell vacuole ruptures, releasing proteases, which then degrade cellular components [47]. Protease-mediated PCD is essential for plant hypersensitive responses [reviewed by Sueldo and van der Hoorn [48]] which limit the spread of pathogens during the biotrophic phase of infection.
The protease inhibitor experiments were intended to test the hypothesis that at least some of the differentially expressed peptidases are involved in limiting the latent growth of Lm hyphae -hyphal growth beyond the edge of the visible lesion -in 74-44 BL. Possibly, this could occur through a role of peptidases in PCD. The lack of differences between the treatments were not understood. However, because the protease inhibitors were applied to the surface of inoculated cotyledons, it is possible that they inhibited fungal proteases, which may be, to some extent, required by Lm for infection.
Another possibility is that the applied protease inhibitors were unable to penetrate the cotyledon cuticles and thus failed to interact with plant proteases. It is also possible that proteases do not make a significant contribution to QR to Lm in 74-44 BL; this is supported by the lack of protease-or peptidase-related GO terms in the enriched GO terms (Table 5).
Chlorophyll A-B binding proteins, which are a source of ROS [49], were also upregulated in inoculated 74-44 BL, relative to Westar. ROS, including hydrogen peroxide, can act as pro-PCD signals [reviewed by ]. The conjecture that the upregulated chlorophyll A-B binding proteins are involved in triggering PCD is supported by our findings that, in inoculated 74-44 BL, the area that stained for hydrogen peroxide was similar to that in Westar, at 5 and 7 dpi, despite the area colonized by Lm hyphae being smaller in 74-44 BL than in Westar ( Fig. 4 and 5). This may indicate that hydrogen peroxide is produced in 74-44 BL beyond the hyphal front, contributing to the restriction to the hyphal growth. In contrast, Lm hyphae grew beyond the zone of hydrogen peroxide production in Westar at 5 and 7 dpi; it appears that ROS production was not able to catch up with the Lm hyphal growth. At 9 and 11 dpi, however, the area occupied by hyphae and ROS were not different. Potentially the more rapid production of hydrogen peroxide in 74-44 BL, relative to Lm hyphal spreading, is able to prevent the biotrophic growth seen in Westar. Other DEGs that are related to the photosynthetic process include BnaA03g11710D, the putative thiazole biosynthetic enzyme and/or ribulose-1,5-bisphosphate synthetase. Thiazole is a precursor of vitamin B1, or thiamine, which can activate plant defenses [51]. Thus, it is reasonable to speculate that increased vitamin B1 biosynthesis may also contribute to the Lm resistance displayed by 74-44 BL. The fact that Ahn et al. [52] and Boubakri et al. [53] found a relationship between vitamin B1-induced disease resistance and hydrogen peroxide suggests a potential link between overregulation of BnaA03g11710D and increased hydrogen peroxide production, relative to the area colonized by Lm hyphae in 74-44 BL (Fig. 4 and 5). Additionally, three GO terms linked to hydrogen peroxide were identified by the GO enrichment analysis of the DEGs upregulated in Lm inoculated 74-44 BL, as compared to Westar (Table 5); these are consistent with the suggested roles of hydrogen peroxide in the QR response of 74-44 BL to Lm.
One of the DEGs upregulated in Lm-inoculated Westar relative to inoculated 74-44 BL (but not differentially expressed between any other pair of treatments) is a heat shock protein 70 which could be linked to the upregulation of Bax inhibitor-1. Qi et al. [54] noted that overexpression of a heat shock protein 70 inhibited PCD induced by hydrogen peroxide; this finding is consistent with our observation that hydrogen peroxide was produced in a larger area, relative to Lm hyphal colonization, in 74-44 BL cotyledons ( Fig. 4 and 5).
The DEGs involved in endomembrane trafficking, such as small GTPases, sec23/sec24, sec61 and Furthermore, Bax inhibitor-1 is ER localized [56], suggesting that it could also be linked to the ER to/from Golgi vesicle traffic. The UPR occurs in the ER and is, as the name implies, linked to improperly folded proteins. If stress to the ER is severe enough, the UPR can induce PCD [reviewed by Cui et al. [57]]. Hence DEGs with potential roles in protein folding that are more highly expressed in Westar than 74-44 BL (both inoculated), such as the putative peptidyl-prolyl cis-trans isomerase BnaC03g44640D, could also be linked to PCD. The fact that differences were not observed in genomic DNA degradation could suggest that the UPR might be only a signaling mechanism related to QR, and not necessarily PCD. A role for the UPR in canola resistance to Lm is consistent with the findings of Arrano-Salinas et al. [58] documenting a link between the UPR and plant immunity. Indeed, the UPR could also be related to ROS production [59]. Further research into the potential roles of the UPR, endomembrane dynamics and ROS production in plant defenses are merited for a better understanding of QR to blackleg of canola.

Conclusion
QR observed in 74-44 BL cotyledons involves restricting tissue colonization by Lm. This is likely due, at least in part, to the production of ROS beyond the pathogen hyphal growth. ROS may be linked to signal transduction and endomembrane vesicle trafficking. The ability of QR to reduce the growth of Lm hyphae in cotyledons can be significant because it may limit the pathogen movement from infected leaves into the stem, where the most damaging form of the disease takes place. Further research is needed to clarify the molecular and cellular mechanisms involved. Such work, in conjunction with exploration of putative modes of action of QR in other canola cultivars and life stages, could help facilitate judicious use of QR-carrier canola for improved management of blackleg in Canada and around the world.

Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Author Contributions
The work was conducted at the Saskatoon Research and Development Centre, Agriculture and Agri-Food Canada. MH and GP conceived the study and designed the experiments. MH carried out the majority of the lab, growth chamber, greenhouse and microscopy work. She also conducted the bioinformatics and statistical analysis and wrote the paper. CZ transformed Lm with GFP and carried out the analysis of genomic DNA fragmentation and GO enrichment analysis. GP and CZ provided input into the data interpretation, composition of the manuscript and editing the manuscript.

Funding
This study was supported by Canola Agronomic Research Program Project No. CARP 2015-12 entitled "Understanding the mechanisms for race-specific and non-specific resistance for effective use of cultivar resistance against blackleg of canola in Western Canada" administered by Canola Council of Canada.   Tables   Table 1 Differentially expressed genes (DEGs) in Leptosphaeria maculans that are more highly expressed in L. maculans inoculated 74-44 BL than in inoculated Westar.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.