Arabidopsis late blight: infection of a nonhost plant by Albugo laibachii enables full colonization by Phytophthora infestans

Summary The oomycete pathogen Phytophthora infestans causes potato late blight, and as a potato and tomato specialist pathogen, is seemingly poorly adapted to infect plants outside the Solanaceae. Here, we report the unexpected finding that P. infestans can infect Arabidopsis thaliana when another oomycete pathogen, Albugo laibachii, has colonized the host plant. The behaviour and speed of P. infestans infection in Arabidopsis pre‐infected with A. laibachii resemble P. infestans infection of susceptible potato plants. Transcriptional profiling of P. infestans genes during infection revealed a significant overlap in the sets of secreted‐protein genes that are induced in P. infestans upon colonization of potato and susceptible Arabidopsis, suggesting major similarities in P. infestans gene expression dynamics on the two plant species. Furthermore, we found haustoria of A. laibachii and P. infestans within the same Arabidopsis cells. This Arabidopsis—A. laibachii—P. infestans tripartite interaction opens up various possibilities to dissect the molecular mechanisms of P. infestans infection and the processes occurring in co‐infected Arabidopsis cells.


Introduction
Plants have evolved diverse and effective mechanisms to protect against attack by microbial pathogens. Indeed, a central tenet of plant pathology is that resistance is the rule and disease the exception (Briggs, 1995). Although broad host-range pathogens do occur, most plant pathogens are adapted to a limited number of taxonomically related host species and cause disease on only a few host plants. Those pathogens may not fare well on plants unrelated to their hosts because of adaptive evolution, which tends to drive organisms towards specialization, for example through the accumulation of mutations that enhance virulence on one host but impair it on another (Borhan et al., 2008;Dong et al., 2014;Dong et al., 2015;Ma et al., 2010;Raffaele et al., 2010;Tosa et al., 2006). In addition, nonhost resistance and species-specific resistance serve to restrict the host range of plant pathogens (Lee et al., 2014;Schulze-Lefert and Panstruga, 2011;Senthil-Kumar and Mysore, 2013). Physical barriers, such as fortified cell walls and a waxy cuticle, production of antimicrobial secondary metabolites, and cell-autonomous immunity all contribute to nonhost resistance (Bettgenhaeuser et al., 2014;Fellbrich et al., 2002;Miedes et al., 2014;Piasecka et al., 2015). Further, cell autonomous immunity is multi-layered, involving pre-invasive defences as well as cell surface and cytoplasmic immune receptors that perceive pathogens (Dodds and Rathjen, 2010;Win et al., 2012). Thus, a pathogen's ability to colonize a certain plant species includes its capacity to suppress or tolerate host immunity.
The oomycete plant pathogens comprise numerous hostspecific species (Fawke et al., 2015;Kamoun et al., 2015;Lamour and Kamoun, 2009;Thines and Kamoun, 2010). These filamentous microorganisms are some of the most destructive plant pathogens and remain persistent threats to both farmed and native plants (Akrofi et al., 2015;Enzenbacher et al., 2015;Hansen, 2015;Roy, 2015). For example, the Irish potato famine pathogen Phytophthora infestans, the causal agent of late blight, recurrently endangers global food security (Fisher et al., 2012;Fry et al., 2015). P. infestans is thought to have a relatively narrow host range, infecting a few wild Solanum species in their native habitats of central Mexico and the Andes, as well as cultivated potato and tomato in most regions where these crops are grown (Fry et al., 2009;Goss et al., 2014;Grunwald and Flier, 2005). In compatible hosts, P. infestans proliferates an extensive host-intercellular hyphal network and projects digit-like haustoria into single host cells (Blackwell, 1953). P. infestans can also infect other solanaceous plants, such as petunia and the experimental host Nicotiana benthamiana (Becktell et al., 2006;Chaparro-Garcia et al., 2011). However, this pathogen is not known to complete its full infection cycle on plants outside the Solanaceae. For example, the model plant Arabidopsis thaliana, a member of the Brassicaceae family, is fully resistant to P. infestans and is considered a nonhost (Huitema et al., 2003;Lipka et al., 2005;Stein et al., 2006;Vleeshouwers et al., 2000).
On Arabidopsis leaves, as on other nonhost plants such as tobacco and parsley, P. infestans cysts germinate, form appressoria, and directly penetrate epidermal cells to form infection vesicles and occasionally secondary hyphae (Colon et al., 1992;Huitema et al., 2003;Naton et al., 1996;Schmelzer et al., 1995;Vleeshouwers et al., 2000). However, this early interaction is followed by the hypersensitive response, a localized cell death reaction of plants that restricts the spread of the pathogen (Huitema et al., 2003;Vleeshouwers et al., 2000). In the Arabidopsis pen2 mutant, which is deficient in the hydrolysis of 4-methoxyindol-3ylmethylglucosinolate (4MO-I3M) into antimicrobial metabolites, the frequency of P. infestans penetration of epidermal cells increases, resulting in markedly enhanced hypersensitive cell death (Westphal et al., 2008). However, P. infestans does not complete its full infection cycle on pen2 mutants or pen2 mutants combined with mutations in other defencerelated genes (Kopischke et al., 2013;Lipka et al., 2005;Westphal et al., 2008). In these mutants, P. infestans hyphae fail to colonize the Arabidopsis mesophyll to the extent seen in compatible interactions and do not develop haustoria, the specialized hyphal extensions that project into host cells and are thought to be sites where the pathogen secretes virulence proteins (effectors) (Schornack et al., 2010;Whisson et al., 2007). To date, there are no published reports of Arabidopsis mutants that are fully deficient in nonhost resistance to P. infestans, and thus enable extensive biotrophic colonization and sporulation of this pathogen (Geissler et al., 2015;Stegmann et al., 2013).
One oomycete pathogen that can infect A. thaliana is A. laibachii, one of several specialist Albugo species that cause white blister rust disease Kemen et al., 2011) by extensively colonizing host-intercellular spaces, projection of knob-like haustoria into host cells (Soylu, 2004), and by forming visible blister-like dispersal pustules of sporangia on the lower side of leaves. Albugo spp. are obligate biotrophic parasites that are phylogenetically distinct from other oomycetes, such as Phytophthora, and thus have independently evolved the ability to colonize plants (Kemen and Jones, 2012;Thines and Kamoun, 2010). Albugo are widespread as endophytes in asymptomatic natural populations of Brassicaceae and likely influence the biology and ecology of their host species (Ploch and Thines, 2011). Remarkably, Albugo can suppress host immunity to enable colonization by other races of pathogens and subsequent genetic exchange between specialized genotypes with nonoverlapping host ranges (McMullan et al., 2015). Prior infection by Albugo enhances susceptibility to plant pathogens such as downy and powdery mildews (Bains and Jhooty, 1985;Cooper et al., 2008). For instance, pre-infection with A. laibachii enables avirulent races of the Arabidopsis downy mildew Hyaloperonospora arabidopsidis to grow and sporulate on resistant Arabidopsis accessions . A. laibachii suppresses the runaway cell death phenotype of the Arabidopsis lesion simulating disease1 mutant, further supporting the view that this pathogen is an effective suppressor of plant immunity . The mechanisms by which Albugo spp. suppress immunity remain unknown, but probably involve suites of effector genes like those identified in the Albugo candida and A. laibachii genomes (Kemen et al., 2011;Links et al., 2011).
Here, we aimed to determine the degree to which A. laibachii would enable maladapted pathogens to colonize Arabidopsis. Pre-infection with A. laibachii did not alter resistance of Arabidopsis to the Asian soybean rust pathogen (Phakopsora pachyrhizi) or the powdery mildew pathogen (Blumeria graminis f. sp. hordei (Bgh)). However, we discovered that pre-infection with A. laibachii enables the potato pathogen P. infestans to fully colonize and sporulate on Arabidopsis, a plant that is considered to be a nonhost of this Solanaceae specialist. Our results show that when exposed to A. laibachii colonized tissues, P. infestans carries the potential to infect other plant species outside its natural host spectrum employing a conserved set of transcriptionally induced effector encoding genes. The interaction of Arabidopsis-A. laibachii-P. infestans will be an excellent model to examine how co-infection of host cells enables infection by P. infestans.

A. laibachii infection enables P. infestans colonization of the nonhost plant Arabidopsis
Previous work indicated that the potato late blight pathogen P. infestans can penetrate epidermal cells of its nonhost Arabidopsis resulting in hypersensitive cell death but little ingress beyond the infection site (Huitema et al., 2003;Vleeshouwers et al., 2000). To determine the extent to which A. laibachii alters the interaction between Arabidopsis and P. infestans, we carried out serial inoculations with the two pathogens. First, we infected rosette leaves of 5-week-old Arabidopsis Col-0 with spores of A. laibachii strain Nc14 (Kemen et al., 2011). Successful infections were identified based on the formation of white sporangiophores on the abaxial side of rosette leaves 10 days after inoculation. At that stage, we detached the infected leaves, inoculated them with zoospores of P. infestans 88069 and monitored symptom development (Fig. 1A-B). Within 5 days after inoculation with P. infestans, we observed water-soaked tissue, necrosis, and ultimately sporulation in co-infected leaves (Fig. 1B). As controls we also applied P. infestans zoospores to uninfected leaves of A. thaliana Col-0, and also monitored mock-and A. laibachii-inoculated leaves ( Fig. 1 A-B). No necrosis was observed in these negative controls ( Fig. 1 A-B). To further investigate the degree to which P. infestans colonizes pre-infected Arabidopsis leaves, we repeated the experiment with P. infestans 88069td, a transgenic strain that expresses the cytoplasmic red fluorescent protein (RFP) marker tandem dimer, and monitored pathogen ingress by microscopy. This revealed an extensive network of red fluorescent P. infestans hyphae and unlabelled A. laibachii hyphae in the co-infected leaves that extended to most of the leaf within just 3 days after P. infestans inoculation and sharply contrasted with the P. infestans-only treatment ( Fig. 2 A-B). We also repeated the experiment with whole plants to ensure that the observed effect was not an artifact of the detached leaf assay. Here also, P. infestans triggered severe disease symptoms and formed an extensive hyphal network only in the mixed-infection leaves ( Fig. 1 C-E).
Next, we quantified pathogen biomass during infection using kinetic PCR as previously described (Judelson and Tooley, 2000;Mauch et al., 2009) (Fig. 3). We amplified the P. infestans gene PiO8 to estimate relative levels of P. infestans DNA in infected plant tissue and observed a continuous increase over time in Arabidopsis leaves preinfected with A. laibachii (Fig. 3). We quantified A. laibachii biomass upon P. infestans co-infection by using read data from our RNA sequencing experiment. We found numbers of reads matching to established constitutively expressed A. laibachii genes do not change upon infection with P. infestans ( Fig. S1) suggesting that co-colonization does not have a detrimental effect on A. laibachii colonization. Overall, the pathology, microscopy, and molecular biology experiments indicate that P. infestans becomes able to fully A. Control leaves (Mock) or leaves from A. thaliana Col-0 plants pre-infected with A. laibachii were detached and droplets of water (H 2 0) or P. infestans spore solution were applied to their abaxial sides and incubated for 4 days in high humidity. B. A close-up of (A) reveals P. infestans sporulation (arrowheads) as a dense cover of leaves pre-infected by A. laibachii only. B. Albugo laibachii enables P. infestans to colonize leaves infected on a whole plant. A. thaliana Col-0 plants treated with water (H 2 0) or pre-infected with A. laibachii were inoculated with droplets of water (H 2 0) or P. infestans spore solution and incubated in high humidity. Macroscopic observations of disease symptoms on whole plants at 3 days post inoculation. D. A close-up of (C) reveals P. infestans disease symptoms only on leaves pre-colonized by A. laibachii (right panel). E. The extent of P. infestans 88069td hyphal colonization (under RFP illumination) was assessed 3 days post inoculation using epifluorescence microscopy. Scale bar = 250 μm. All experiments were performed at least twice with similar results. colonize nonhost Arabidopsis plants upon pre-infection of those plants with A. laibachii (Figs 1-3, Fig. S1).

Cellular dynamics of P. infestans colonization of pre-infected Arabidopsis
To study the interaction between P. infestans and preinfected Arabidopsis in more detail, we performed confocal microscopy on leaves inoculated with P. infestans strain 88069td. We conducted side-by-side comparisons of the subcellular interactions of P. infestans in A. laibachii-and mock-infected leaves. In both cases, we observed germinated P. infestans cysts on the leaf surface as well as appressoria (Fig. 4A, Fig. 4B) and infection vesicles within the plant epidermal cells (Fig. 4C, Fig. 4D). The difference between the two treatments became apparent at 1 day post infection (dpi) with the activation of host cell death (the hypersensitive response, HR) visible through accumulation of autofluorescent material in epidermal cell walls at sites of attempted infection by P. infestans in mock-treated leaves only ( Fig. 4E-F). Arabidopsis pre-infected with A. laibachii did not display an HR at sites of penetration of P. infestans (Fig. 4G, Fig. 4H). To independently validate these data, we inoculated P. infestans strain 88069td on leaves that were mock treated or pre-infected with A. laibachii, and then stained the leaves to quantify dead cells and monitor the invasion process at two different time points (6 h post infection, hpi, and 24 hpi) (Fig. 4I). This again confirmed that penetration of P. infestans was not associated with the HR in samples that were pre-infected with A. laibachii, at both 6 and 24 hpi (Fig. 4I).
The ingress of P. infestans beyond its infection site became apparent starting at 36 hpi (1.5 dpi) in the A. laibachii pre-infected leaves, with intercellular hyphae spreading from the penetration site ( Fig. S2). In contrast to mock-treated samples, the hyphae extended at 3 dpi to colonize the mesophyll and most of the leaf (Fig. S2). Branching hyphae with narrow, digit-like haustoria expanded from the site of penetration to neighbouring cells through the intercellular space ( Fig. S3). Starting at 3 dpi, the mycelium developed sporangiophores that released numerous sporangia to produce zoospores (Fig. S4). Thus, the P. infestans colonization of Arabidopsis pre-infected with A. laibachii resembles, in behaviour and speed, the P. infestans infection reported on susceptible potatoes (Vleeshouwers et al., 2000).
A. laibachii and P. infestans haustoria within A single Arabidopsis cell A. laibachii forms haustoria in Arabidopsis cells (Caillaud et al., 2012). Because we observed the formation of haustoria by P. infestans in Arabidopsis pre-infected with A. Abaxial sides of control leaves of A. thaliana Col-0 (left column) and leaves pre-infected with A. laibachii (right column) have been infected with red fluorescent P. infestans 88069td. The extent of A. laibachii sporulation (under GFP illumination visible as green autofluorescent pustules, upper row) and P. infestans hyphal colonization (visible under RFP illumination as red hyphal network, middle row) was assessed 3 days post inoculation using epifluorescence microscopy. Bottom row represents merged fluorescence pictures. B. Abaxial sides of co-infected leaves at 2 dpi exhibiting dual colonization of P. infestans (in red) and by A. laibachii hyphae (not fluorescently labelled, indicated by yellow arrowheads) within the same area. All experiments were performed twice with similar results. Abbreviations: ps: pustules, h: hyphae; s: spores, gc: germinating cyst. Scale bars = 250 μm (A) or 50 μm (B).
A. laibachii, we searched for cells that harboured haustoria of both oomycetes. We recorded numerous events where single or multiple digit-like P. infestans haustoria cooccurred with multiple knob-like A. laibachii haustoria in the same Arabidopsis cells (Fig. 5). In 45% of all assessed Arabidopsis mesophyll and epidermal cells with P. infestans haustoria we observed A. laibachii haustoria within the same confocal plane (N obs = 17; 2 independent experiments). Thus, haustorium formation by A. laibachii or P. infestans does not trigger processes that prevent secondary penetration by another species. This observation will enable us to study how focal redirection of cellular compartments is affected by secondary penetration and how the two microbial pathogens vary in recruiting plant secretory processes to their haustoria.
In planta expression dynamics of P. infestans secreted protein genes are similar on Arabidopsis and potato Expression analyses have identified a significant set of P. infestans effector genes, which are transcriptionally induced during biotrophy in host-plant infections (Cooke et al., 2012;Haas et al., 2009;Pais et al., 2013). These studies have been limited to infections of potato and tomato, which both belong to the nightshade family (Solanaceae). To test whether the induced effector gene set is different in Arabidopsis pre-infected with A. laibachii, we collected A. laibachii-infected and mock-infected Arabidopsis leaves at different time points following application of zoospores of P. infestans strain 06_3928A (13_A2 clonal lineage, (Cooke et al., 2012)). To compare sets of differentially regulated P. infestans genes in Arabidopsis with those differentially regulated in potato, we also infected and harvested potato leaves. Extracted RNA from all samples was subjected to Illumina RNA-seq.
We found that during colonization of potato, the steadystate transcript levels of 10 698 P. infestans genes were significantly altered. Of those, 7118 transcripts were also altered the same direction in A. laibachii pre-infected Arabidopsis. In contrast, 776 transcripts were exclusively altered in the P. infestans-A. laibachii-Arabidopsis interaction (see Supplementary Table 1 for details).
We next examined changes in transcripts encoding secreted proteins and found 196 induced sequences, of which 136 (66%) were shared, 40 were uniquely induced in Arabidopsis/A. laibachii, and 20 uniquely induced in potato (Fig. S5A, Fig. 6A). We found a strong correlation between Arabidopsis/A. laibachii and potato in the degree of gene expression induction both at 2 and 3 dpi with P. infestans (Fig.  6B, Fig. S5). Out of a total of 96 induced effector gene transcripts, a common set of 78 (81%) were induced in both plant species, whereas 12 and 6 effector transcripts were induced in a host-specific manner during colonization of Arabidopsis and potato respectively. Seven RXLR effector genes with known avirulence activity in specific potato cultivars were similarly induced in both host species (Fig. 6C). In summary, we conclude that the induction of secreted protein genes of P. infestans during colonization of potato and Arabidopsis/A. laibachii leaves do not greatly differ.
Arabidopsis leaves pre-infected with A. laibachii do not become susceptible to barley powdery mildew fungus or Asian soybean rust fungus To determine the degree to which the effect of A. laibachii on P. infestans extends to other maladapted pathogens, Arabidopsis leaves pre-infected with A. laibachii were inoculated with the fungal pathogens B. graminis f. sp. hordei (Bgh) and P. pachyrhizi, the agents of barley powdery mildew and Asian soybean rust respectively. In both cases we observed no alteration of the interactions (Figs S6 and S7). Both of these fungal pathogens failed to penetrate leaves of both mock-and A. laibachii pre-infected Arabidopsis plants.

Discussion
In this study, we demonstrated that the potato blight pathogen P. infestans becomes capable of colonizing Arabidopsis Five-week old leaves of A. thaliana Col-0 pre-infected with A. laibachii were detached and drop-inoculated with a zoospore suspension of P. infestans isolate 06_3928A or mock-treated with water applied to their abaxial sides and incubated for 4 days under high humidity. DNA was extracted at 0, 1, 2, 3, and 4 days post inoculation and used for quantitative PCR (qPCR) for PiO8 with gene-specific primers for P. infestans. Pathogen DNA levels were normalized to the Arabidopsis SAND gene (At2g28390) and the relative amount of Pi08 was normalized to the DNA level in mock-inoculated samples. Data are representative of one biological replicate with three technical replicates of qPCR reaction. Bars represent ratio between mean normalized expression of the infected samples with P. infestans and A. laibachii and the mock-treated sample with A. laibachii and water (calibrator) (Mean ± SE). Data was analysed using repeated measures two way-ANOVA (P < 0.0001). Letters indicate significant results of Fisher's LSD post-hoc test. Experiment was performed twice with similar results. when this nonhost plant is pre-infected by the obligate parasite A. laibachii. This is surprising, given that P. infestans is a Solanaceae specialist that is seemingly maladapted to plants from other botanical families. We took advantage of this tripartite interaction to perform comprehensive cellular and molecular analyses. On A. laibachii-infected Arabidopsis, P. infestans goes through its full infection cycle to a degree that has not been observed to date with pre-and post-invasive mutants (Geissler et al., 2015;Kobae et al., 2006;Lipka et al., 2005;Stegmann et al., 2013;Stein et al., 2006;Westphal et al., 2008). This includes the formation of haustoria, rapid hyphal proliferation, and profuse sporulation (Figs 1-5, Figs S1-S4). Expression dynamics of P. infestans genes encoding secreted proteins and effectors on susceptible (i.e., pre-infected) Arabidopsis were generally similar to those on potato, indicating that this pathogen colonizes pre-infected Arabidopsis in a similar manner as it colonizes its usual host plant (Fig. 6, Fig. S5B). In this study we used two different P. infestans isolates (88069td and 06_3928A) and obtained convergent results from transcriptome analysis and cell biological studies. Therefore, colonization of Albugo-infected Arabidopsis is not strain specific.
P. infestans is a hemibiotroph and as such will initially colonize tissues biotrophically but will subsequently kill the tissue and feed on the remains. The initial biotrophic colonization including formation of haustoria is essential to colonization of host plants. We observed haustoria during colonization of A. laibachii-infected Arabidopsis. Furthermore, autofluorescence monitoring using fluorescence microscopy and cell death trypan blue staining (Fig. 4) show presence of cell death only upon P. infestans singular infection, but not upon A. laibachii/P. infestans co-infection. The induced expression of effector encoding genes previously associated with biotrophy in the P. infestans-potato host system lends further support to an initial biotrophic growth. In conclusion, there are no data supporting an immediate necrotrophy when applying P. infestans spores to A. laibachii-preinfected tissues. Instead, P. infestans exerts a hemibiotrophic lifestyle on potato as well as on A. laibachii colonized Arabidopsis.
The finding that P. infestans can fully colonize immunosuppressed plants distantly related to its hosts indicates that pathogen host range may not be fully determined by a lack of essential factors in the nonhost, one of several resistance mechanisms generally thought to determine host specificity (Agrios, 2005). Indeed, there is little evidence that nonhost resistance results primarily from the absence of taxonspecific factors in the plant. For example, Garber's nutritional theory, which postulates that resistant plants provide a 'nutritional environment that is inadequate for a parasite' (Garber, 1956), has received little support over the years. By contrast, a greater understanding of the versatility and efficacy of the plant immune system has led to the view that active pre-and post-invasive defences play a preponderant role in protecting most plants against most pathogens, and therefore in ultimately delimiting pathogen host range (Dodds and Rathjen, 2010;Jones and Dangl, 2006).
Our findings are consistent with the evolutionary history of the P. infestans lineage, which reflects significant plasticity in host range. This lineage, also known as clade 1c, consists of a tightknit group of closely related species that have specialized on host plants from four different botanical families as a consequence of a series of host jumps (Dong et al., 2014;Grunwald and Flier, 2005;Raffaele et al., 2010). This indicates that on a macroevolutionary scale, the P. infestans lineage has the capacity to generate variants that can infect divergent host plants (Dong et al., 2015). The split between P. infestans and its sister species P. mirabilis is estimated to have occurred relatively recently~1300 years ago , providing some indication of the frequency of host jumps within the clade 1c lineage.
A. laibachii converts Arabidopsis into a fully susceptible host of P. infestans to a degree that has not been observed to date with genetic mutants. The pen2 and pen3 mutants, which are deficient in penetration resistance, display enhanced responses to P. infestans, exhibiting a macroscopically visible hypersensitive cell death that results from increased frequency of epidermal cell penetration (Kobae et al., 2006;Lipka et al., 2005;Stein et al., 2006). However, the extent to which Arabidopsis penetration resistance to P. infestans is effective at stopping pathogen ingress is debatable given that penetration events can also be observed on wild-type Arabidopsis (Huitema et al., 2003;Vleeshouwers et al., 2000). In this study, we confirmed that penetration of Arabidopsis epidermal cells by P. infestans germinated cysts is commonly observed on wild-type Arabidopsis (Fig. 4). Thus, although pen mutants enable increased plant cell penetration, pre-invasive barriers do not fully block P. infestans infection, given that infection vesicles can be readily observed on mock-treated wild-type Arabidopsis at 16 hpi (Fig. 4). This view is consistent with the dramatic effect we observed on plants pre-infected with A. laibachii, which did not display P. infestans-triggered hypersensitivity probably as a consequence of post-invasive immunosuppression. Consistent with a post-invasive effect, A. laibachii did not alter Arabidopsis resistance to pathogens such as barley powdery mildew (Fig. S6) and Asian soybean rust fungi (Fig. S7), which cannot penetrate wild-type Arabidopsis cells, in sharp contrast to P. infestans (Figs S2 and S3).
In P. infestans, as in many other filamentous pathogens, the expression of a subset of genes, notably secreted protein genes, is markedly induced during host infection (Cooke et al., 2012;Haas et al., 2009;Jupe et al., 2013;Pais et al., 2013). The mechanisms that underpin host signal perception by these pathogens, and the nature of these signals, remain largely unknown. We noted that the set of P. infestans effector genes induced on susceptible Arabidopsis largely (A) (B) (C) Fig. 6. Similar sets of effectors are induced during P. infestans colonization of potato (Solanum tuberosum) and Arabidopsis pre-infected with A. laibachii. A. Numbers of commonly and uniquely induced genes encoding secreted P. infestans proteins and effectors (a subset of the secreted proteins). B. Dot blot comparing the transcript levels of P. infestans effector-encoding genes between S. tuberosum and Arabidopsis pre-infected with A. laibachii at 2 (left panel) and 3 days post infection (right panel). C. Gene expression intensities relative to the average expression intensity in media (Rye sucrose) are shown for genes encoding avirulence proteins (Gene IDs in parentheses) during the interaction of P. infestans with A. laibachii pre-infected Arabidopsis leaves (left panel) and S. tuberosum leaves (right panel) induced at 2 dpi and 3 dpi. Genes encoding ubiquitin ligases, Elongation factor 2, and Actin are shown as uninduced controls. All expression intensities are log2 transformed. overlaps with the genes induced in the host plant potato (Fig. 6). Patterns of effector gene expression displayed similar dynamics on both plants, with a peak during the biotrophic phase at 2 dpi. These results indicate that it is unlikely that P. infestans perceives a host-specific plant signal to trigger in planta gene induction. One possibility is that as the pathogen progresses from host cell penetration to intercellular hyphal growth to haustorium formation, it undergoes a developmental programme that regulates gene expression. Thines (2014) recently put forward the theory that Albugoinfected plants could serve as a bridge that enables other oomycetes to shift from one host plant to another. Indeed, repeated cycles of co-infection may facilitate the selection of genotypes of the maladapted pathogen that are virulent on the nonhost, eventually leading to a host jump. This scenario may have occurred with downy mildew species of the genus Hyaloperonospora, which tend to share Brassicaceae hosts with Albugo spp. (Thines, 2014). However, the degree to which Albugo has affected the ecological diversification of P. infestans and possibly other Phytophthora is unclear. First, it is not known whether the two pathogens are sympatric in central and south America, the natural geographic range of P. infestans and its sister species (Goss et al., 2014;Grunwald and Flier, 2005). Second, unlike P. infestans, most Phytophthora spp. are soil pathogens that do not spread aerially and are thus unlikely to colonize Albugo-infected leaves. Nonetheless, the possibility that biotic agents, such as A. laibachii, have facilitated host jumps in the P. infestans lineage should not be disregarded and deserves to be studied, for example by genome sequencing of environmental leaf samples. Our study further highlights the importance of studying multitrophic interactions in order to fully understand the biology and ecology of plant pathogens (Kemen, 2014).
Few diseases rival the effect of P. infestans on humankind (Fisher et al., 2012;Yoshida et al., 2013). Long after it triggered the Irish potato famine, this pathogen is still regarded as a threat to global food security and is an active subject of research . To date, P. infestans research has focused mainly on its interaction with Solanacaeous plants. Little progress has been achieved using model systems such as A. thaliana, and work on Arabidopsis-P. infestans has been limited to studies of nonhost resistance (Huitema et al., 2003;Kopischke et al., 2013;Lipka et al., 2005). Other Phytophthora spp., e.g. P. brassicae, P. cinnamomi, P. parasitica, and P. capsici, have been shown to infect Arabidopsis but they have been exploited to a lesser extent in research (Roetschi et al., 2001;Robinson and Cahill, 2003;Belhaj et al., 2009;Wang et al., 2011;Wang et al., 2013). The Arabidopsis-A. laibachii-P. infestans tripartite interaction opens up several new avenues of research: (i) to address the genetic diversity of Arabidopsis resistance towards P. infestans, (ii) to define the degree to which Albugo spp. have influenced the ecological diversification of P. infestans and enabled host jumps throughout evolution, and (iii) to dissect the molecular mechanisms, and focal retargeting of plant secretory pathways of co-infected host cells, a situation that is likely to occur frequently under natural conditions.

Biological material
A. thaliana plants were grown on an 'Arabidopsis mix' (600 L F2 compost, 100 L grit, 200 g Intercept insecticide) in a controlled environment room (CER) with a 10 h day and a 14 h night photoperiod and at a constant temperature of 22°C. A. thaliana Col-0 ecotype was used for all experiments.
P. infestans isolates and 88069 expressing a cytosolic tandem RFP protein (88069td) and P. infestans strain 06_3928A (13_A2 clonal lineage) were cultured on rye sucrose agar at 18°C in the dark as described earlier (Chaparro-Garcia et al., 2011;Cooke et al., 2012). A. laibachii strain Nc14 was used in pre-infection experiments in this study (Kemen et al., 2011). This strain was maintained on the A. thaliana Col-5 line containing multiple insertions of the RPW8 powdery mildew resistance gene (Col-gl RPW8.1 RPW8.2) (Xiao et al., 2001). The infected plants were kept overnight in a cold room (5°C) then transferred to a growth cabinet under 10 h light and 14 h dark cycles with a 21°C day and 14°C night temperature as described (Kemen et al., 2011). Besides P. infestans and A. laibachii we used two obligate fungal parasites: B. graminis f. hordei CH4.8 (IPKBgh) and P. pachyrhizi isolate PPUFV02. A summary of fungal isolates used in this study and how they were maintained is provided in Supplementary Table 2.

Sequential infection assays
All infection assays were performed on four-or five-week-old Arabidopsis plants of ecotype Col-0. Plants were pre-inoculated with a zoospore suspension of A. laibachii (7.5 × 10 5 spores/ml) obtained from zoosporangia released from 14-day-old treated Col-gl RPW8. ) droplet was applied to the abaxial side of the leaf. Leaves were incubated on a wet paper towel in 100% relative humidity conditions with a 14 h/10 h day/night photoperiod and at a constant temperature of 18°C.
Co-infection assays with powdery mildew pathogen (B. graminis f. sp. hordei isolate CH 4.8) were performed on detached Arabidopsis leaves. Three-centimetre leaf strips were cut from the cotyledon or first leaf of the barley cultivar and used as a control.
Leaves were placed into agar plates containing 100 mg/l benzimidazole. Powdery mildew spores were collected from the barleyinfected leaves on a piece of paper. Infection was made in a settling tower by tapping and blowing the inoculum. Plates were allowed to settle for 10 min after infection in the tower before incubation in a growth cabinet at 15°C (16 h light/8 h dark with 18°C light/13°C dark) (Brown and Wolfe, 1990).
Co-infection assays with the Asian soybean rust were performed on detached leaves with P. pachyrhizi isolate PPUFV02 as described (Langenbach et al., 2013). Briefly, uredospores from P. pachyrhizi-infected soybean leaves were collected at 14 days-post inoculation (dpi), suspended in 0.01% (v/v) Tween-20 at 1 mg/ml and used for inoculation. Spore suspension of P. pachyrhizi was sprayed on Arabidopsis leaves until the droplets covered the whole leaf surface. To allow fungal spore germination, infected leaves were maintained in moist conditions (100% humidity) and in the dark for the first 24 hpi.

Cytological analysis of infected material
Arabidopsis leaves infected with the red fluorescent P. infestans Identical microscope settings were applied to all individuals.
To quantify the HR cell death response in infected samples, leaves were stained with lactoglycerol-trypan blue and washed in chloral hydrate as described earlier . Specimens were mounted on microscope slides and analysed with a Leica DM2700 M microscope (Leica Microsystems).
Powdery mildew structures were stained with lactoglycerol-trypan blue as described earlier (Vogel and Somerville, 2000). Briefly, excised leaves were destained in ethanol overnight than washed thoroughly with in water and placed in lactoglycerol (1:1:1 lactic acid: glycerol:water). Specimens were mounted on microscope slides with a few drops of 0.1% lactoglycerol-trypan blue staining on top.
Fungal structures were imaged with a Leica DM2700 M microscope (Leica Microsystems).
Asian soybean rust-infected tissues were stained as described earlier in Ayliffe et al. (2011). Briefly, Arabidopsis leaf tissue was placed in 1 M KOH, then neutralized in 50 mM Tris, pH 7.0. The leaf was then stained with wheat germ agglutinin conjugated to fluorescein isothiocyanate (WGA-FITC, Sigma-Aldrich, UK) at 20 μg/ml. Specimens were mounted on a microscope slide and analysed with a Leica DM6000B/TCS SP5 confocal microscope (Leica Microsystems) with an excitation wavelength for GFP of 458 nm.
All microscopy images acquired for the various infections were analysed by using the Leica LAS AF software, ImageJ (2.0) and Adobe PHOTOSHOP CS5 (12.0).
SYBR Green (Qiagen, UK) was used as fluorescent reporter dye to amplify the PiO8 gene and was normalized to the Arabidopsis SAND gene (At2g28390) which was amplified with the following primer pairs SAND-F (5′-AACTCTATGCAGCATTTGATCCACT-3′) and SAND-R (5′-TGATTGCATATCTTTATCGCCATC-3′)  value (Efficiency) of each reaction was calculated. The ratio between P. infestans and Arabidopsis genomic DNA was calculated using the REST method as described (Pfaffl et al., 2002).

Statistical analysis
All data were analysed using the Prism software version 6.01 (GraphPad Software, USA . Leaf discs were collected at 2 and 3 days post inoculation (dpi) using a cork borer No. 4. Infected leaf samples were ground in liquid nitrogen until a fine powder was obtained and stored at À80°C prior to RNA extraction. We used the RNeasy Plant Mini Kit (Qiagen,Cat No. 74904), following the manufacturer's instructions, to extract total RNA for all samples. cDNA libraries were prepared from total RNA using the TruSeq RNA sample prep kit v2 (Cat No. RS-122-2001). Library quality was confirmed before sequencing using the Agilent 2100 Bioanalyzer (Agilent Technologies). Sequencing was carried out using an Illumina Genome Analyzer II (Illumina Inc) with TruSeq Cluster generation kit v5  and TruSeq Sequencing kit v5 . We performed read quality control by removing reads containing Ns and reads with abnormal read length (other than 76 bases) using FASTX-Toolkit version 0.0.13 (http://hannonlab.cshl.edu/fastx_toolkit). Total reads (76 bp paired-end) that passed the parameters mentioned above for quality control were used for downstream analyses (Supplementary Table   3 Table 3). A two-stage analysis of the pathogen reads was applied to rescue multi-mapped or ambiguous reads that cannot be uniquely assigned to groups of genes. First, we generated Reads Per Kilo Base per Million (RPKM) values for each gene by using the htseq-count script that is part of the HTSeq python module (Anders et al., 2015). Next, we rescued reads that were enriched for gene families using multi-map group (MMG) approach and customized perl scripts (Robert and Watson, 2015). In brief, we allocated multi-mapped reads based on probability of multi-mapped reads derived from particular locus that was calculated from RPKM, and then estimated final RPKM according to a published method (Mortazavi et al., 2008). The adjusted-RPKM values of all reads after rescues were transformed into Log2 fold values by dividing the RPKM data to the RPKM values from mycelium of P. infestans isolate 06_3928A (Wagner et al., 2012). In planta-induced genes exhibiting at least two-fold gene induction between averaged media and infected sample (at 2 and/or 3 dpi) were considered induced during infection. Log2 values were loaded in Mev4_8 version 10.2 TM4 microarray software suite (Saeed et al., 2003) and analysed using Col-0 precolonized with A. laibachii was inoculated with red fluorescent P. infestans 88069td and imaged at 3 days post inoculation with confocal laser scanning microscopy. The upper panel shows confocal micrographs of hyphal extension and sporulation of P. infestans in A. laibachii pre-treated Arabidopsis leaves. The lower panel is a close-up of the region highlighted by the dotted square in the upper panel and shows P. infestans emerging sporangiophores from the leaf surface, giving rise to lemonshaped zoosporangia. Experiment was performed twice with similar results. Scale bar = 100 μm (upper panel) or 10 μm (lower panel). Fig. S5. P. infestans genes encoding secreted proteins during infection of potato and Arabidopsis precolonized with A. laibachii. The heat map (A) illustrates 325 genes encoding secreted proteins; these genes were induced at least 2-fold during the interaction of P. infestans with S. tuberosum leaves at 2 dpi and 3 dpi and during the interaction with Arabidopsis leaves colonized with A. laibachii at 1, 2, and 3 dpi. These genes were mean-centred and hierarchically clustered by Euclidean distance. (B) Expression dynamics of selected P. infestans AVR genes and constitutively expressed control genes at 1, 2, and 3 days post inoculations (dpi). Fig. S6. A. laibachii does not enable the nonhost powdery mildew pathogen to infect nonhost Arabidopsis. A. thaliana Col-0 mock-treated (control) or precolonized with A. laibachii was inoculated with Blumeria graminis f.sp hordei (Bgh) isolate CH4.8. The susceptible Barley cv. Golden Promise was used as a control for the infection. (A) Macroscopic phenotype of disease symptoms two weeks post inoculation with Bgh isolate CH4.8. (B) Maximum projection of images produced by light microscopy from 14 Z-stacks from infected tissues. Micrographs show fungal structures 2 weeks post infection in both control (right panel) and samples pre-colonized with A. laibachii (left panel) and reveal that the fungus was stopped at the penetration stage in both interactions. Experiment was performed twice with similar results. Abbreviations: c: conidium, pgt: primary germ tube, sgt: secondary germ tube, ap: appressorium, pa: papillae, ha: haustoria of A. laibachii, #: conidiospores of A. laibachii. Scale bar = 10 μm. Fig. S7. A. laibachii does not enable the nonhost Asian soybean rust to infect nonhost Arabidopsis Five-week-old A. thaliana Col-0 mock-treated (control) or precolonized with A. laibachii was inoculated with spore suspension of Phakopsora pachyrhizi isolate PPUFV02 and incubated in high humidity. Pathogen structures were visualized with confocal laser scanning microscopy under GFP illumination at 6 days post inoculation. The micrographs show uredisniospores germinating on the leaf surface, producing an appressorium with no further growth in both control (upper panel) and pre-colonized samples with A. laibachii (middle and lower panels). All experiments were performed twice with similar results. Abbreviations: u: urediniospore, ap: appressorium, st: stomata, #: conidiospores of A. laibachii. Scale bar = 10 μm. Table S1. List of expressed genes in both potato/P. infestans and A. thaliana/A. laibachii-P. infestans interactions. The file compiles three different worksheets containing lists (list 1-3) of genes of P. infestans and their expression profiles during infection in potato (List 1) and/or in A. thaliana preinfected with A. laibachii (List 2-3). List 1 comprises 10 698 coding genes of P. infestans on potato. List 2 comprises 7118 coding genes that are expressed in A. thaliana preinfected with A. laibachii only when compared to potato-P. infestans interaction. List 3 is a set of 325 in planta-induced genes encoding secreted proteins in both treatments (potato and pre-infected A. thaliana with A. laibachii). Abbreviations: GSR: gene sparse region, GDR: gene dense region, InBW: in between in between gene sparse and gene dense, NA: not applicable. If an NA is present in a column with annotations, this refers to no annotation to that type or description. If NA is present in a column with expression data this refers to a lower expression than 2-fold (or log2 < 1) and in consequence it was not considered as in planta induced for that particular gene. Table S2. Summary of pathogen isolates and media or susceptible plants used for maintaining pathogens. Table S3. Alignment statistics of P. infestans isolate 06_3928A RNA sequences in infected materials. Pair-end reads of 06_3928A isolate were aligned to the reference genome strain T30-4 with the TopHat software package.