The Histological, Effectoromic, and Transcriptomic Analyses of Solanum pinnatisectum Reveal an Upregulation of Multiple NBS-LRR Genes Suppressing Phytophthora infestans Infection

Utilization of disease resistance components from wild potatoes is a promising and sustainable approach to control Phytophthora blight. Here, we combined avirulence (Avr) genes screen with RNA-seq analysis to discover the potential mechanism of resistance in Mexican wild potato species, Solanum pinnatisectum. Histological characterization displayed that hyphal expansion was significantly restricted in epidermal cells and mesophyll cell death was predominant, indicating that a typical defense response was initiated in S. pinnatisectum. Inoculation of S. pinnatisectum with diverse Phytophthora infestans isolates showed distinct resistance patterns, suggesting that S. pinnatisectum has complex genetic resistance to most of the prevalent races of P. infestans in northwestern China. Further analysis by Avr gene screens and comparative transcriptomic profiling revealed the presence and upregulation of multiple plant NBS-LRR genes corresponding to biotic stresses. Six NBS-LRR alleles of R1, R2, R3a, R3b, R4, and Rpi-smira2 were detected, and over 60% of the 112 detected NLR proteins were significantly induced in S. pinnatisectum. On the contrary, despite the expression of the Rpi-blb1, Rpi-vnt1, and Rpi-smira1 alleles, fewer NLR proteins were expressed in susceptible Solanum cardophyllum. Thus, the enriched NLR genes in S. pinnatisectum make it an ideal genetic resource for the discovery and deployment of resistance genes for potato breeding.


Introduction
Potatoes, one of the most important food crops, have been affected by late blight disease for nearly 180 years [1]. To decrease the yield losses caused by Phytophthora infestans, cultivars of Solanum tuberosum expressing diverse resistance (R) genes have been developed and extensively applied in potato growing areas worldwide. Many typical R genes, such as R1 to R11 from Solanum demissum, were initially demonstrated to be successful in improving disease-resistant commercial potatoes. However, the R-gene-insensitive strains of P. infestans frequently emerged after large-scale application of resistant cultivars harboring major R genes [2,3]. For instance, more aggressive P. infestans isolates escaping Int. J. Mol. Sci. 2020, 21, 3211 3 of 17 virulent to all tested potato R-gene differentials (R1-R11). The detached leaves of the susceptible species S. cardiophyllum and resistant species S. pinnatisectum were inoculated with 10 3 zoospores. As shown in Figure 1a, only restricted water-soaked lesions developed at 72 h post inoculation (hpi) on S. pinnatisectum leaves. Well-expanded diseased areas with fluffy hyphae and a large number of sporangia were observed on the leaf surface of S. cardiophyllum (Figure 1a). The relative biomass of P. infestans measured at 72 hpi via qPCR was more than 2-fold greater in diseased tissue on S. cardiophyllum than that detected on leaves of S. pinnatisectum (Figure 1b). At the later stage of infection (108 hpi), the lesions expanded to the whole S. cardiophyllum leaves, but lesions on S. pinnatisectum were only slightly larger than at 72 hpi ( Figure 1c). Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 3 of 17 to all tested potato R-gene differentials (R1-R11). The detached leaves of the susceptible species S. cardiophyllum and resistant species S. pinnatisectum were inoculated with 10 3 zoospores. As shown in Figure 1a, only restricted water-soaked lesions developed at 72 h post inoculation (hpi) on S. pinnatisectum leaves. Well-expanded diseased areas with fluffy hyphae and a large number of sporangia were observed on the leaf surface of S. cardiophyllum (Figure 1a). The relative biomass of P. infestans measured at 72 hpi via qPCR was more than 2-fold greater in diseased tissue on S. cardiophyllum than that detected on leaves of S. pinnatisectum (Figure 1b). At the later stage of infection (108 hpi), the lesions expanded to the whole S. cardiophyllum leaves, but lesions on S. pinnatisectum were only slightly larger than at 72 hpi ( Figure 1c).  (Each error bar represents the mean ± SD, ** and *** indicate statistical significance where p < 0.05 and p < 0.001 using t test, respectively).

Histological Characteristics of S. pinnatisectum Infected by P. infestans
To investigate the histological characteristics of early interactions between P. infestans and S. pinnatisectum or S. cardiophyllum, detached leaves were drop inoculated with zoospores and collected at 3, 6, 9, and 12 hpi, respectively. Within 3 hpi, no significant difference was observed between S. pinnatisectum and the susceptible control S. cardiophyllum. More than 80% of the cysts germinated on the leaf surface of both species, and most of them produced an appressorium at the tips of the germ tubes (Table 1, Figure 2a,b; both Table 1 and Figure 2 were modified from Cao [24]). Invasive hyphae were observed as well, which were predominant at the anticlinal walls between the epidermal cells (Figure 2a,b). By 6 hpi, infection vesicles were observed beneath appressoria in the epidermal cells of both S. pinnatisectum and the susceptible control (Figure 2c,d). The intercellular hyphae were generated and extended intercellularly among spongy cells, although penetration into the spongy tissue of S. pinnatisectum was occasionally visible (Table 1)  Hpi refers to hours post-inoculation; GC abbreviate for short germinated cysts; A is the abbreviation for appressoria formed by cysts observed; HE is short for hyphae infected in epidermis; HS is short for hyphae infected in spongy tissue; HP is short for hyphae infected in palisade tissue.
Phenotype differences between S. pinnatisectum and S. cardiophyllum became more evident at 9 hpi. In S. pinnatisectum, more than 90% of the infection hyphae were confined to the site of the initial penetration of the epidermal cell, while in S. cardiophyllum, the infection hyphae extended rapidly, with more than 40% of the hyphae reaching spongy tissue, and the haustoria were visible (Table 1, Figure 2e,f). In S. pinnatisectum, the infection was still restricted to the epidermal cells by 12 hpi, and some epidermal cells and mesophyll cells beneath the infection site were necrotic (Figure 2h). By this time point, nearly 60% of the hyphae spread into the spongy tissue of S. cardiophyllum, and about 25% of the hyphae had invaded into the palisade tissue but necrosis was rarely observed (Table 1, Figure 2g). These results provided preliminary evidence that a typical defense response rather than cell wall-associated penetration resistance was initiated in S. pinnatisectum upon P. infestans inoculation. Based on the above microscopic observations, we determined 3, 9, and 12 hpi as time points to investigate the transcriptional dynamics of S. pinnatisectum compared to S. cardiophyllum.

S. pinnatisectum Reveals Complicated Pathotypes Based on Multiple R Genes
To dissect the genetic resistance of S. pinnatisectum, the detached leaves were subjected to inoculation with a panel of 12 P. infestans isolates displaying SSR-based genotype variations [5] and R-gene-dependent pathotype diversity ( Table 2). The phenotypes of both S. pinnatisectum and S. cardiophyllum to each isolate were evaluated macroscopically and qualitatively based on a hypersensitive response (HR) or expanding sporulating lesions. As shown in Figure 3a, isolate 10, 11, and 12, and H 2 O control failed to generate any visible lesions, whereas isolate 1, 2, and 5 produced many tiny brown spots on S. pinnatisectum. Lesions caused by isolate 3, 4, and 6 were restricted to inoculation sites, but isolate 8 and 9 were more virulent on the inoculation sites of S. pinnatisectum ( Figure 3a). Based on the pathotypes of isolate 8 and 9, functional homologs of R1, R2, R5, R8, R9, R10, and R11 may not be present in S. pinnatisectum or are not sufficient to trigger defense responses. On the contrary, only isolate 7 was avirulent on S. cardiophyllum (Figure 3d). Ten other isolates, except isolate 5, were more aggressive on the susceptible control S. cardiophyllum (Figure 3b-e). Interestingly, hypervirulent isolate 12, escaping the recognition of all 11 R genes S. demissum, and the isolate 6, which is avirulent on most of the R1-R11 differential hosts, produced similar pathotypes on S. pinnatisectum and S. cardiophyllum (Figure 3a,c). Moreover, we noticed that isolates 8 and 9, showing avirulence on R5, R8, and R9 differential hosts, were virulent on S. pinnatisectum. However, isolate 6, showing avirulence on R5, R8, and R9 differential host as well, was avirulent on S. pinnatisectum. Altogether, the distinct resistance patterns observed in the detached-leaf assay illustrate that the resistance machinery in S. pinnatisectum is complicated and no single R gene of R1-R11 determines the avirulence phenotype.

RxLR Effector Screening Reveals Different R Protein Composition in S. pinnatisectum and S. cardiophyllum
Ten cloned notable RxLR effector genes (Avr genes) and two Avr-R gene pairs were delivered into S. pinnatisectum or S. cardiophyllum leaves by agroinfiltration. Based on the development of cell death symptoms, it is clear that S. pinnatisectum and S. cardiophyllum possess distinct recognition spectra of the Avr genes tested (Figure 4, Supplementary Table S1). Of the 10 tested Avr genes, 6 of them, Avr1, Avr2, Avr3aKI, Avr3b, Avr4, and Avrsmira2, triggered hypersensitive responses in S. pinnatisectum (Figure 4a,b), whereas in S. cardiophyllum, cell death was observed on leaves expressing Avrblb2, Avrsmira1, and Avrvnt1 (Figure 4c). The two potato species have distinct matching R genes, all of which belong to typical NBS-LRR resistance genes. The higher number of R genes in S. pinnatisectum may be responsible for its strong resistance to diverse isolates of P. infestans. Avr3aKI and Avr3a-R3a both triggered cell death in S. pinnatisectum, but functional R3a was absent in S. cardiophyllum. It is remarkable that neither Avrvnt1 nor the Avrvnt1/Rpivnt1 pair produced cell death in S. pinnatisectum, indicating that the presence of a specific interacting protein rather than a functional homolog of Rpi-vnt1 is needed for the initiation of HR.
all of which belong to typical NBS-LRR resistance genes. The higher number of R genes in S. pinnatisectum may be responsible for its strong resistance to diverse isolates of P. infestans. Avr3aKI and Avr3a-R3a both triggered cell death in S. pinnatisectum, but functional R3a was absent in S. cardiophyllum. It is remarkable that neither Avrvnt1 nor the Avrvnt1/Rpivnt1 pair produced cell death in S. pinnatisectum, indicating that the presence of a specific interacting protein rather than a functional homolog of Rpi-vnt1 is needed for the initiation of HR. More interestingly, this effector-based avirulence pattern does not perfectly match the inoculation pathotypes. For instance, R1 and R2 functional alleles were detected via the effector infiltration assay in S. pinnatisectum, and some Avr1or Avr2-expressing isolates still caused infection whereas some Avr1and/or Avr2-expressing isolates did not (Figure 4a, Table 2). Similarly, even though Rpi-blb2, Rpi-smira1, and Rpi-vnt1 are expressed in S. cardiophyllum (Figure 4c) and the expression of some cognate Avr genes was validated in several tested isolates [25], 11 out of the 12 tested P. infestans isolates were more aggressive on this wild potato species (Table 2). These results raise the possibility that pathogens deliver multiple effectors to suppress effector-triggered immunity (ETI) and induce potato susceptibility.

Transcriptome Dynamics of NLR Genes in Susceptible and Resistant Potatoes against P. infestans
According to the previous histological observations in susceptible and resistant cultivars upon P. infestans challenge, four key time points were established at the early stage of infection. Inoculated leaves of S. cardiophyllum or S. pinnatisectum were collected at 6, 9, and 12 hpi, and samples at 0 hpi were used as the control, respectively. Illumina sequencing was conducted to generate transcriptome profiles of both the pathogen and potato over the infection time course. A summary of the number of potato reads after sequence filtering to remove short-length and low-quality sequences is shown in Supplementary Table S2. Approximately 31-70% of the sequences from each sample were mapped to the S. tuberosum genome.
To characterize the molecular basis of the defense mechanism induced after pathogen infection, we identified the NLR genes (see the materials and methods) from the potato genome and investigated their expression. A total of 123 NLR genes were annotated. Of these, 112 have at least one read in at least one sample. Clustering of the expression profiling of the NLR genes revealed that they could be divided into two groups related to the resistant (Sp) and susceptible (Sc) potato cultivar, respectively ( Figure 5). Previous studies showed that Rpi1 and Rpi2 have been mapped to chromosome 7 of S. pinnatisectum [20]. However, according to our RNAseq data, only two NLRs, PGSC0003DMG401030700 and PGSC0003DMG400017317, were dramatically upregulated in the chromosome 7 region of S. pinnatisectum and S. cardophyllum, respectively. In general, over 60% of NLRs were induced and exhibited higher expression levels in S. pinnatisectum but not in S. cardiophyllum after P. infestans infection. In particular, R1 homologue (PGSC0003DMG402004578), R2 homolgue (PGSC0003DMG400025259), and R3a-like genes (PGSC0003DMG400009455, PGSC0003DMG400018570, and PGSC0003DMG400027377) were identified in the upregulated NLRs of the Sp group. Thus, the massive expression of NLR genes in S. pinnatisectum is probably responsible for the resistant phenotype of S. pinnatisectum. While most of the NLR genes coinciding with S. pinnatisectum resistance are highly expressed at 12 hpi, some of them are expressed immediately upon infection ( Figure 5), indicating that S. pinnatisectum may recruit different NLRs at different stages to deal with the pathogen attack.

Discussion
Compared to other wild potato species, very few novel R genes or functional homologues have been identified from S. pinnatisectum in the last two decades, but this Mexican wild potato species

Discussion
Compared to other wild potato species, very few novel R genes or functional homologues have been identified from S. pinnatisectum in the last two decades, but this Mexican wild potato species has been proposed to contain potential qualitative and quantitative resistance against P. infestans. Kuhl et al. [18] characterized 13 accessions of S. pinnatisectum by means of P. infestans zoospore inoculation, and more than 10 of them showed resistance, including accession PI275233. A large proportion of 44 S. pinnatisectum accessions were reported to be resistant to P. infestans in laboratory assays and in field sprays [8]. In agreement with the above results, our studies verified that S. pinnatisectum displayed considerable resistance against multiple P. infestans isolates, which are characterized by different SSR genotypes and diverse pathotypes [5]. Especially, those predominant isolates, which are virulent to the R1-R11 potato differential set, failed to infect this wild potato species. On the contrary, these supervirulent isolates induced rapidly expanding water-soaked lesions on susceptible cultivars of S. cardiophyllum. Remarkable resistance or tolerance against P. infestans was observed from 3 to 72 hpi, and much lower amounts of inoculum proliferated in the diseased tissue of S. pinnatisectum.
The interaction between Avr and R protein triggered immediate and strong hypersensitive responses (HRs) in plant cells, which led to lesion-free phenotypes or could be visible as localized cell death [26]. The responses of plants to pathogen stress were more obvious and pervasive in mesophyll cells within 9 to 12 hpi but were rare in the susceptible control, S. cardiophyllum. Necrosis was visible in the infected mesophyll cells of S. pinnatisectum at 12 hpi and later, indicating that the HR was induced upon infection. As a result, expansion of P. infestans was restricted to the epidermis, consistent with other resistant germplasm of potato [27]. Significant variations of histological characteristics in the infection processes between S. pinnatisectum and the susceptible control S. cardiophyllum also illustrates that defense responses are different from penetration resistance as cysts normally geminated and penetrated into epidermal cells before 6 hpi. Thus, the R gene-mediated ETI probably plays a key role in disease resistance and tolerance under P. infestans challenge. The RNA-seq data verified this hypothesis by revealing an upregulation of many ETI components.
To sketch the genetic resistance of S. pinnatisectum, we first revaluated the late blight resistance of S. pinnatisectum by using different P. infestans isolates with distinct pathotypes. Surprisingly, only two isolates (No. 8 and 9) could readily infect S. pinnatisectum, indicating that the expression of Avr1 and Avr2 did not affect the infection. Supervirulent dominant isolates (No. 12) rarely infected this resistant cultivar but displayed high aggressiveness on S. cardiophyllum. Pathotype association analysis suggests that multiple genes govern the resistance. The use of P. infestans strains could be challenging, because multiple R genes in resistant cultivars can mask recognition specificities. Many resistance breeding studies have employed Phytophthora RxLR effectors rather than diverse P. infestans isolates as markers to mine novel R genes or dissect functional R gene alleles [28]. To better discriminate the specific R gene activities, in this research, we also took advantage of the RxLR effector screen and RNA-seq analysis and discovered a distinct group of NLR genes that are responsible for the remarkable resistance of S. pinnatisectum.
The functional homologues of qualitative R genes R1, R2, R3a, R3b, and R4 and quantitative R gene Rpi-smira2 could be accurately detected using Avr1, Avr2, Avr3aKI, Avr3b, Avr4, and Avrsmira2 in S. pinnatisectum. The pathogenic variations of different Avr1and Avr2-expressing isolates suggest that other potential Avr genes are involved in R gene recognition or ETI suppression. Such a special case is that the loss of resistance of Rpi-blb1 (RB) is caused by ETI suppressing the variant of Avrblb1 [29]. Thus, the detection of mutations in cloned Avr probe genes may not be sufficient for a virulent profile prediction of filed isolates. This is different from the scenario of P. sojae, in which the presence of functional Avr genes in selected isolates is highly consistent with their avirulent phenotype [30]. Moreover, co-expression of Avrvnt1 and Rpi-vnt1 or single Avrvnt1 failed to induce HR in S. pinnatisectum, suggesting that specific interacting proteins are needed for Avrvnt1-Rpi-vnt1 recognition. As Gao C et al. pointed out [31], glycerate 3-kinase (GLYK) is involved in activation of of Rpi-vnt1. Similarly, Avrblb1 and Rpi-blb1 only triggered typical HR in N. benthamiana but not in Arabidopsis thaliana, indicating that species-specific interactors are involved in cell death induction [32]. According to the guard model and the decoy model, the Avr-R protein recognition specificity would depend on secondary molecules. For example, BSL1 was initially identified as a key chaperone for PiAvr2 and R2 interaction, and then was predicted to regulate multiple R gene-mediated immunity responses [33]. As a result, future studies should not only seek identification of novel genes conferring resistance but also patterns of defense genes activated in ETI.
None of those six R genes exist in S. cardiophyllum, but three other R genes (Rpi-blb2, Rpi-vnt1, and Rpi-smira1) were detected in this susceptible control by transient expression of cognate Avr genes. These R genes did not provide adequate resistance for this cultivar and resulted in high susceptibility. Currently, for late-blight management, breeders prefer to pyramid multiple R genes in one cultivar to avoid the rapid defeat of R genes [11]. Here, we have an opposite example, as although Rpi-blb2 and Rpi-vnt1 were documented as a broad-spectrum resistance gene from S. bulbocastanum and S. venturi [34,35], the natural combination of Rpi-blb2, Rpi-vnt1, and Rpi-smira1 in S. cardiophyllum changed its susceptibility against most isolates used in our detached-leaf inoculation assay. Moreover, the mutation or silencing of cognate RXLR effectors may not be responsible for the loss of resistance, since the transcripts of conserved Avrvnt1, Avrblb2, and Avrsmira1 were detected in at least one of isolate 1, 6, or 12 [25]. Thus, more effort is necessary for the construction of desired long-term stable resistant cultivars.
Despite only part of the known Avr genes being employed in the screening, the results suggest that more active R genes were clustered in S. pinnatisectum. It is also reported that even defeated R genes in wild potatoes may lead to a delay in infection. Considering that all tested local isolates of P. infestans contain truncated Avr4 [25], our findings suggest that the remarkable resistance of S. pinnatisectum is probably due to pyramiding of R1, R2, R3a, R3b, and Rpi-smira2. Of course, genetic validations are required to test whether the desired resistance can be attributed to these functional alleles. Then, the major genes for late blight resistance can be cis-transformed into potato cultivars. In addition, the degree of resistance is dependent on the genotype of the P. infestans population, and monitoring for virulence to specific R genes in the local P. infestans population can assist the deployment of R genes [36]. Once the transgenic techniques are broadly accepted, it could accelerate genetic modifications with R genes.
The RNA-seq data revealed that a massive number of immune-associated NLR genes, including a distinct NLR gene group, were dramatically upregulated in S. pinnatisectum. Despite the low similarity between different orthologs and some species-specific NLR genes being absent in our mapping process, the RNA profile analysis validated the previous effector screen and demonstrated the expression of multiple NLR genes. Considering that most of the known resistant genes belong to the NBS-LRR superfamily with a 3-kb average length [17], we mapped RNAseq reads to 123 full-length NLR genes with an entire NB-ARC and LRR domain. Similarly, Andolfo et al. employed cDNA RENseq to identify 124 and 221 full-length NB-LRR in two tomatoes [37]. Furthermore, taking advantage of the RENseq technique to enrich pattern recognition receptors (PRRs) [38], NLR and other ETI components could provide a better understanding of the broad and durable resistance mechanisms in S. pinnatisectum.
Our preliminary investigation of the resistance mechanisms in S. pinnatisectum provides an opportunity to explore potential novel ETI components and the effective combination of R genes for late-blight resistance breeding. Indeed, the employment of genetic engineering into late-blight resistance breeding programs should speed up the race against P. infestans. However, rational deployment of effective R genes and gene pyramiding can extend the durability of such resistance and offers potato growers an environmentally friendly alternative for late-blight management.

Plant and Pathogen Materials
The diploid potato species, S. pinnatisectum (Accession PI275233) and S. cardiophyllum (Accession PI186548), which are resistant or susceptible to P. infestans, respectively, were used in this study. Both species were propagated in vitro in sterile jars containing Murashige and Skoog medium supplemented with sucrose and grown for about 6 weeks after transplanting into nutrient soil at 20-25 • C under a 14 h:10 h day/night photoperiod.
All P. infestans isolates were previously described [5,25] and were routinely maintained on RSA (rye sucrose agar) medium. The numerous sporangia collected from 7-to 10-day cultures were chilled with cold sterilized water at 4 • C for about 1 h to release zoospores. The zoospores were filtered through one layer of Miracloth and the concentration was adjusted to 1 × 10 5 zoospores/mL. A highly aggressive isolate of P. infestans Pi21336, which overcomes the R1-R11 genes of potato, was used for inoculation of S. pinnatisectum and S. cardiophyllum in the histological and transcriptomic analyses. Additionally, 12 other isolates with complex virulence spectra ( Table 2) were also used for detecting the genetic resistance of S. pinnatisectum. The pathogenicity of the 12 P. infestans isolates on S. pinnatisectum and S. cardiophyllum was evaluated 3-4 days post-inoculation (dpi). The inoculation site for the 12 isolates on each leaflet was randomized. Three leaflets of S. pinnatisectum and 12 leaflets of S. cardiophyllum were used for each repetition and the experiment was repeated twice.

Lesion Measurement and Biomass Analysis
Pictures of diseased leaves were taken at 4 dpi, and the relative lesion sizes were measured with ImageJ software. For the qPCR analyses, samples (20 µL) were prepared by mixing 1 µL (5 ng) of gDNA and 19 µL of 2 × SYBR Green mix (Bimake, Houston, TX, USA) with the appropriate primers (Table S3) added to a final concentration of 5 µM each and water. PCRs were performed in triplicate for each biological sample, using the default relative quantification program of CFX connect Real-Time PCR System (Bio-Rad, Hercules, CA, USA). Ct values were determined with the included Bio-Rad software. The detection of real-time PCR products, calculations, and statistical analysis were performed as previously described [39].

Microscopic Characterization of S. pinnatisectum Infected by P. infestans
Detached-leaf inoculations were conducted as previously described [25] with some modifications. Leaflets were detached from robust growing plants in the greenhouse, placed on water-saturated filter paper in a tray, and inoculated on the abaxial side of both S. pinnatisectum and S. cardiophyllum with 10-µL droplets containing approximately 10 3 zoospores. The inoculated leaves were kept in moisture trays sealed with plastic wrap at 18 • C in the dark for the first 12 h, and then the trays were incubated at 18 • C under a photoperiod of 16 h light and 8 h darkness. For microscopic observations, samples were obtained at 3, 6, 9, 12, and 72 hpi. Samples were stained with trypan blue staining following the protocol described by Wang et al. [40]. The experiment was repeated at least three times.
Agroinfiltration was performed as previously described [25]. Briefly, A. tumefaciens strains were grown to an optical density at 600 nm of 0.2, and leaf panels of 6-to 8-week-old potato plants were infiltrated with the A. tumefaciens suspensions. The position for infiltration was randomized.