Arabidopsis plants were grown in short day (8 h light/16 h dark), 12 h light/12 h dark and long days (16 h light/8 h dark) as described in Engelsdorf et al. (2013). All mutants, AtSWEET11-GFP and AtSWEET12-GFP reporter plants used in this study are described in Chen et al. (2010, 2012), except for sid2-3 (SALK_042603) that is described in Gross et al. (2006). pAtSWEET11:AtSWEET11-YFP and pAtSWEET12:AtSWEET12-YFP reporter constructs were made by cloning pAtSWEET11:AtSWEET11 and pAtSWEET12:AtSWEET12, respectively, from the donor vector pDONR221-f1 (Chen et al., 2012) into the pEG-TW Gateway vector (Chen et al., 2012; Fan et al., 2014) by LR clonase. Transgenic plants were produced as described previously (Chen et al., 2012) and were a generous gift of Li-Qing Chen and Wolf Frommer (Carnegie Institution for Science, Stanford, CA, United States). All mutants and transformants were in the Col-0 background.

Colletotrichum higginsianum wild type isolate MAFF 305635 (Ministry of Agriculture, Forestry and Fisheries, Japan) and CIH1-mCherry overexpressors were used. The CIH1-mCherry expressing strain was generated by Agrobacterium mediated transformation of MAFF 305635 using pCIH1-mCherry. For the construction of pCIH1-mCherry, ChCIH1 (CH063_13023) including 1012 bp upstream sequences was fused in frame to an mCherry gene in pSL1180 and subsequently cloned into the single PmeI site of binary vector pPN (Korn et al., 2015). Prior to the use in infection assays, fungal strains were grown on oat meal agar plates [5% (w/v) shredded oat meal, 1.2% (w/v) agar] for 7 days at 22°C under illumination to promote conidia formation. Inoculation of leaves from 5-week-old Arabidopsis plants with Ch was performed by evenly spraying a conidia suspension at a titer of 2 × 10⁶ conidia/ml onto the leaves as described by Voll et al. (2012). For droplet infections, 5 μl of a 2 × 10⁶ conidia/ml suspension were placed on the leaf surface and plants were further treated as in spray infections.

To assess early in planta establishment of Ch between 1.5 and 2.5 days post infection (dpi), the developmental stages of about 200 to 400 fungal structures were analyzed in whole-leaf mounts after lactophenol-trypan blue staining as described in Koch and Slusarenko (1990) with minor modifications. After boiling freshly sampled leaves in trypan blue staining solution for 1 min, the infected leaves were kept in the staining solution for 24 h and were then transferred to chloral hydrate [250% (w/v)] for destaining on the following day. Microscopy was performed on a Leica DMR microscope (Bensheim, Germany) with DIC optics.

For later infection stages, quantification of the relative genomic DNA content of Ch was performed according to Engelsdorf et al. (2013).

Localization of AtSWEET11-YFP and AtSWEET12-YFP in Arabidopsis leaves was performed on a Leica TCS SP5 confocal laser scanning microscope (Leica, Wetzlar, Germany). After excitation with an argon laser at 514 nm, YFP fluorescence was recorded with a photomultiplier tube (PMT) detector between 525 and 560 nm. Ch expressing CIH1-mCherry was visualized after excitation with a DPS5 laser at 561 nm and emission PMT detector settings between 570 and 630 nm. Chlorophyll autofluorescence was detected between 657 and 721 nm after excitation with an argon laser.

Localization of YFP signals on the whole leaf scale was performed using a Leica MZ16F binocular (Leica, Wetzlar, Germany) with the corresponding YFP band pass filter set for excitation (510/20 nm) and emission (560/40 nm).

Extraction and quantitation of free SA, SAG in leaf samples was performed as described (Voll et al., 2012) with minor modifications reported in Engelsdorf et al. (2013).

Snap-frozen Arabidopsis leaves were extracted twice with 80% ethanol and soluble sugar contents were assayed in a coupled enzymatic assay using a microtiter plate reader as described by Voll et al. (2003). Starch measurements from ethanol insoluble leaf material was performed as described by Voll et al. (2003).

For transcriptome analysis on Agilent Arabidopsis V4 4x44K microarrays (Design-ID 021169, Agilent, Waldbronn, Germany), total RNA was extracted from pools of twelve fully expanded leaves per replicate using the RNase-all method (Chomczynski and Sacchi, 1987). Leaf samples of 5-week-old sweet11/sweet12 and wild type Col-0 plants grown in 12 h light (20°C)/12 h dark cycles (18°C) were either taken immediately before the treatment, at 1 h before the end of the light period (0 dpi) and at 2.5 dpi, at 3 h into the light period. At 2.5 dpi, both mock treated and Ch infected samples were taken. Prior to microarray hybridization, RNA quality was assessed by an Agilent 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany). Gene expression and GO term enrichment analysis was performed with GeneSpring V12.6 (Agilent Technologies, Waldbronn, Germany).

Student’s t-tests were performed with SigmaPlot 12 (Systat Software Inc., Chicago, IL, United States) after testing for normality (Shapiro–Wilk test) and equal variance.

It was shown that SWEET sucrose transporters are induced by bacterial effectors in the phloem of rice leaves to establish compatibility with Xoo (Chen et al., 2010; Yuan et al., 2011). In Arabidopsis leaves, various sets of SWEET transporters were found to be induced upon infection with bacterial and fungal pathogens (Chen et al., 2010). Since AtSWEET11 and AtSWEET12 were shown to exert redundant functions in phloem loading of Arabidopsis (Chen et al., 2012), we first investigated if either or both transporters might play a role in the interaction of Arabidopsis with Ch.

Immediate post penetration establishment of Ch was evaluated microscopically in the middle of the biotrophic phase at 2.5 dpi. The relative number of appressoria (A), biotrophic structures (infection vesicles and primary hyphae, PH) and necrotrophic SH was scored in histological specimen and was taken as a measure for the progress of fungal in planta development (Figure 1A). In wild type leaves, around 55% of the assessed appressoria had successfully formed biotrophic hyphae in planta, while only 20% of the penetration attempts on sweet11/sweet12 double mutants had resulted in establishment of the biotrophic phase (Figure 1A). The sweet11 and sweet12 single mutants showed an intermediate phenotype (Figure 1A). Necrotrophic hyphae were only occasionally observed at that time point in all genotypes. Thus, the establishment of biotrophy in planta was significantly delayed in the sweet11/sweet12 double mutant, while early post penetration of Ch was also attenuated in both single mutants, indicating an additive effect of AtSWEET11 and AtSWEET12 deficiency on early post-penetration.

Fungal colonization of the double mutant was reduced by approximately 50% relative to wild type in the early necrotrophic phase at 3.5 dpi (Figure 1B), while no significant differences between wild type and both single mutants were observed in any of seven independent replicate experiments (results of one representative experiment are shown in Figure 1B).

Since deficiency of both transporters showed an additive effect on compatibility, we next assessed if AtSWEET11 and AtSWEET12 are induced by Ch. Transcript profiling in the biotrophic (2 dpi) and necrotrophic infection phase (4 dpi) indicated that none of the two transporters was substantially induced during biotrophic colonization, while only AtSWEET12 got strongly induced on the whole leaf scale during the necrotrophic phase of Ch infection (Table 1).

To assess the local and temporal induction of AtSWEET11 and AtSWEET12 more closely, we performed droplet inoculation with Ch on pAtSWEET11:AtSWEET11-YFP and pAtSWEET12:AtSWEET12-YFP reporter plants. Confirming the microarray data, AtSWEET11-YFP protein was detected in vascular tissue of infected and mock control leaves, but no further induction of AtSWEET11-YFP was observed in Ch infected leaves (Supplementary Figure S1A). In contrast, AtSWEET12-YFP accumulated at Ch infection sites during the biotrophic phase at 2.5 dpi, and strong expression of AtSWEET12-YFP fusion protein was detected around necrotic lesions during the necrotrophic infection phase at 4 dpi (Figure 2A, see Supplementary Figure S1B for images from an independent pAtSWEET12:AtSWEET12-YFP reporter line). In addition, AtSWEET12-YFP accumulation was much stronger in the vasculature of infected compared to control plants at both stages of the infection (Figure 2A). To assess the local induction of AtSWEET12 during the biotrophic phase at the cellular level, we infected the pAtSWEET12:AtSWEET12-YFP reporter lines with a Ch strain expressing a secreted LysM-Domain CIH1-mCherry reporter protein, which accumulates around primary hyphae in planta (Figure 2B and Supplementary Figure S2). Interestingly, we never detected AtSWEET12-YFP around primary hyphae of the pathogen by CLSM. An accumulation of AtSWEET12-YFP was only observed in 9% of the penetrated epidermis cells (Figure 2B). By contrast, neighboring epidermis cells and underlying mesophyll cells showed AtSWEET12-YFP reporter fluorescence in 84 and 61% of all cases, respectively (Figure 2B).

The absence of AtSWEET11-YFP and AtSWEET12-YFP from the interfacial matrix strongly argues against a role of these two transporters in providing sugars to Ch directly at the plant–pathogen interface. However, induction of AtSWEET12 in mesophyll and epidermis cells surrounding the infected cell might enrich the apoplasmic compartment at the infection site with sucrose. Therefore, we investigated major carbohydrate metabolism in control and Ch infected leaves in more detail.

Since the sweet11/sweet12 double mutant exhibited an approximately 50% reduction in sugar exudation and reduced nocturnal starch mobilization (Chen et al., 2012), we expected that soluble sugar levels are constantly elevated in the sweet11/sweet12 double mutant.

To study major carbohydrate contents in more detail, we measured the contents of soluble sugars and starch in leaves during diurnal light/dark cycles in short day (8 h light/16 h dark), 12 h light/12 h dark cycles and long day conditions (16 h light/8 h dark). In all light regimes, the sweet11/sweet12 double mutant showed strongly elevated contents of hexoses, sucrose and starch throughout the diurnal cycle compared to wild type and the single mutants (Figure 3A and Supplementary Figure S3). Throughout the diurnal cycle, sucrose and starch contents remained constantly elevated in the double mutants compared to the other genotypes, while hexose accumulated in the double mutant during the first half of the light period (Figure 3A and Supplementary Figure S3). To assess more closely, if the rate of starch and soluble sugar accumulation (i.e., the content at the end of light minus the content at the end of the preceding dark period) as well as starch and soluble sugar mobilization (i.e., the content at the end of light minus the content at the end of the following dark period) differ between the genotypes, we calculated the turnover of total soluble sugars and starch from the time course data (Figure 3B and Supplementary Figure S3). While the turnover of soluble sugars was significantly increased in the sweet11/sweet12 double mutant under all studied diurnal cycles, the calculated total diurnal carbohydrate accumulation and mobilization were comparable between all other genotypes (Figure 3B and Supplementary Figure S3).

Infected leaves of all genotypes accumulated more hexoses and sucrose than the corresponding mock controls at the end of the light periods at 3 and 4 dpi. Hexose contents in Ch infected leaves were twice as high in the double mutant as compared to the other genotypes at all time points, while sucrose contents were elevated in infected double mutant compared to infected wild type leaves at the end of the subjective light phases at 3 and 4 dpi (Figure 4). Taken together, the progressive accumulation of hexoses and sucrose in the course of Ch infection was much more pronounced in double mutant leaves compared to wild type and single mutants. Leaves of infected single mutants only exhibited a transient increase of hexose and sucrose contents compared to infected wild type leaves at 3 dpi (Figure 4).

The observed elevated levels of soluble sugars, especially hexoses, in mock treated sweet11/sweet12 double mutants – as well as the enhanced accumulation of soluble sugars upon challenge with Ch – may elicit sugar mediated priming of the SA pathway, as described by Linke et al. (2002) and Conrath et al. (2006). Therefore, we monitored contents of free SA as well as conjugated SA glucoside (SAG) in leaves of untreated control plants, mock treated plants and Ch infected plants in 12 h intervals from the time of treatment until 2.5 dpi. At all time points, SA and SAG contents in untreated leaves ranged between 40 and 50 μg⋅m^-2 and 60 and 80 μg⋅m^-2, respectively, and did not differ significantly between genotypes. Both SA and SAG contents rapidly increased at the time of biotrophic establishment of Ch at 2 dpi in all genotypes (Figures 5A,B). While the timing of SA production was comparable among all genotypes, SA and SAG accumulation was increased approximately twofold in sweet11/sweet12 compared to wild type at 2 dpi (Figure 5).

Interestingly, the contents of free SA and SAG were also significantly elevated by around twofold in mock treated sweet11/sweet12 double mutants compared to mock treated wild type, i.e., in plants that were sprayed with water and kept at high humidity (Figure 5C). This might be connected to sugar-mediated priming of the SA pathway in the absence of pathogen. We employed microarray analysis to investigate the response of SA regulated genes in mock treated leaves at 2.5 dpi. In mock treated double mutants, 645 genes were significantly down-regulated and 345 genes were significantly up-regulated compared to mock treated wild type at 2.5 days post treatment (Supplementary Table S1). More than 13% of those genes that were significantly induced in sweet11/sweet12 double mutants, i.e., 134 genes, were associated with plant defense responses according to GO term annotation. GO categories associated with defense, incompatible interactions and SA associated pathways were highly significantly enriched in this gene set (Table 2), corroborating that mock treated sweet11/sweet12 leaves are in a primed state. Concomitantly, Ch infected double mutant leaves exhibited a significant enrichment of the antagonistic JA response among the down-regulated genes compared to infected wild type leaves at 2.5 dpi (corrected p-value 2.57 × 10^-5), indicating a faster establishment of the SA response in sweet11/sweet12 upon Ch challenge. Most obviously, dozens of GO terms associated with photosynthetic and primary metabolism were significantly enriched among the genes induced in Ch infected double mutant leaves compared to Ch infected wild type (Supplementary Table S2), reflecting a reduced degree of infection and, consequently, increased metabolic capacity of the double mutant at 2.5 dpi. In untreated sweet11/sweet12 double mutants only 114 genes were significantly up- and down-regulated, respectively, compared to untreated Col-0 wild type (Supplementary Table S3). An enrichment of defense associated GO categories was absent from these gene sets.

Taken together, sweet11/sweet12 double mutants exhibit enhanced SA accumulation and an induction of SA regulated genes already in mock treated leaves. Since SA mediated responses are reportedly very effective in defense of Ch during biotrophy (O’Connell et al., 2004; Birker et al., 2009), a faster induction of the SA triggered defense response might explain the reduced susceptibility of the double mutant toward Ch. To provide genetic evidence for the role of the SA priming for the diminished susceptibility of double mutants toward Ch, we investigated fungal colonization in sweet11/sweet12/sid2 triple mutants, which lack the committed step of SA biosynthesis, isochorismate synthase, and hence, SA (Wildermuth et al., 2001).

Fungal colonization of wild type Col-0, sweet11, sweet12, and sweet11/sweet12 mutants at 3.5 dpi was comparable to the data shown in Figure 1, while the SA deficient mutant sid2 was hypersusceptible toward Ch (Figure 6). Intriguingly, the sweet11/sweet12/sid2 triple mutant shared the hypersusceptible phenotype with sid2, demonstrating that the diminished susceptibility of the sweet11/sweet12 mutant depends on the SA pathway.

Besides SWEET transporters that have been identified as direct targets of bacterial TAL effectors in rice and cassava (Chu et al., 2006; Yang et al., 2006; Antony et al., 2010; Chen et al., 2010; Liu et al., 2011; Yu et al., 2011; Cohn et al., 2014), a strong transcriptional induction of several SWEET transporters has been reported during the interaction of Arabidopsis and grapevine with systematically diverse biotrophic and necrotrophic pathogens (Chen et al., 2010; Chong et al., 2014), which may indicate that reprogramming of SWEET transporters for the diversion of organic carbon from the host tissue is a recurring motif during microbial attack. Based on the previously observed transcriptional induction seen by Chen et al. (2010), our study was set out to find indication, if reprogramming of the phloem localized Arabidopsis transporters AtSWEET11 and AtSWEET12 might be vital for carbon supply to C. higginsianum.

We observed that early post penetration establishment during the biotrophic phase of Ch was delayed in both, sweet11 and sweet12 single mutants and an additive effect of both mutations in the double mutant was evident. This may indicate a role of AtSWEET11 and AtSWEET12 in nutrient supply to Ch, but it is impossible to determine in vivo carbon transfer rates that could provide direct experimental evidence to verify this hypothesis.

Nevertheless, circumstantial evidence speaks against this hypothesis. While both single mutants showed delayed establishment of Ch in the early biotrophic phase at 2 dpi, only AtSWEET12-YFP reporter protein accumulated around Ch infection sites. Fungal proliferation in the sweet12 single mutant remained unaffected at the end of the biotrophic phase at 3.5 dpi, demonstrating that AtSWEET12 accumulation around the infection site cannot have a substantial influence on the interaction. In contrast, AtSWEET11-YFP expression remained unaffected in Ch infected leaves and was restricted to vascular tissue, making it very unlikely that AtSWEET11 confers major changes in carbohydrate allocation in Ch infected leaves. If AtSWEET11 or AtSWEET12 were targets of fungal effector proteins and were reprogrammed for diversion of assimilates from colonized leaf tissue, a strong induction of the two SWEET fusion proteins at the plant–fungal interface, i.e., the interfacial matrix, around biotrophic hyphae of Ch would have been expected. This was clearly not the case. Recent data indicate that the initial so-called biotrophic phase of Ch rather serves defense suppression than nutrient uptake (O’Connell et al., 2012; Engelsdorf et al., 2013). None of the 93 potential sugar transporters encoded in the Ch genome were highly induced in biotropic hyphae in planta and only 13 sugar transporters were exclusively transcribed in biotrophic hyphae (O’Connell et al., 2012). In addition, C. higginsianum was found to be much more virulent on plants that had been kept in the dark for 60 h and were devoid of free sugars (Engelsdorf et al., 2013). In summary, it is to be expected that the pathogen is largely independent of host sugar transporter expression, if Ch does not rely on carbon provision by host cells during the initial biotrophic phase.

The sweet11/sweet12 double mutant exhibited reduced susceptibility toward Ch both during the biotrophic and in the ensuing necrotrophic colonization phase, which may indicate that AtSWEET11 and AtSWEET12 might act redundantly for the nutrition of Ch. While AtSWEET11 expression could never be detected outside the vasculature, AtSWEET12 was strongly induced in vicinity of Ch infection sites. This argues against a functional redundancy of AtSWEET11 and AtSWEET12 on the local scale at the interaction site, also excluding a major role of the two transporters in carbon provision to the fungal pathogen.

However, both transporters were expressed in the vasculature of Ch infected leaves. Therefore, it seems reasonable to associate the functional redundancy of AtSWEET11 and AtSWEET12 during Ch challenge with their role in phloem loading (Chen et al., 2012). Unlike sweet11 and sweet12 single mutants, the sweet11/sweet12 double mutant exhibited elevated steady state contents of hexoses, sucrose and starch throughout the diurnal cycle, irrespective of the duration of the light phase. This physiological phenotype indicates that AtSWEET11 and AtSWEET12 play redundant roles for major leaf carbohydrate metabolism. Furthermore, AtSWEET11 and AtSWEET12 appear to have an additive effect on compatibility toward Ch (see data in Figure 1). If we assume that their impact on compatibility is connected to their function in carbohydrate export, it seems reasonable that the loss of AtSWEET11 has a more pronounced effect on compatibility than AtSWEET12, since the expression level of AtSWEET11 is more than twofold higher than that of AtSWEET12 (as analyzed with the eFP browser¹).

It has previously been demonstrated that reduced carbohydrate availability in Arabidopsis leaves correlates with susceptibility toward Ch (Engelsdorf et al., 2013). Furthermore, reduced availability of carbohydrates hampered the induction of SA triggered PR genes (Engelsdorf et al., 2013). In turn, diminished susceptibility of the sweet11/sweet12 double mutant toward Ch may simply be explained by a surplus of carbohydrates compared to the other genotypes. Carbohydrates that fail to be exported from the symplast into the apoplasm will likely accumulate in the cytosol (and in the vacuole) of sweet11/sweet12 leaf cells. It has recently been demonstrated that the PAMP-triggered activation of the hexose transporter STP13 reduces compatibility toward Pseudomonas syringae pv. tomato by influencing the allocation of sugars in expense of the apoplasm and in favor of the host cytosol (Yamada et al., 2016). Likewise, these extra carbon reserves in the sweet11/sweet12 double mutant might allow for increased metabolic activity, and hence an enhanced defense response. In support of the proposed connection between elevated carbon status and enhanced defense in the double mutant, we have observed an accumulation of free SA, SA-glucosides (SAG) as well as a highly significant induction of genes involved in the SA mediated defense response not only in Ch infected, but also in mock treated sweet11/sweet12 double mutants relative to wild type. Interestingly, the described stimulation of the SA pathway was absent from untreated double mutants, indicating that elevated contents of soluble sugars are not sufficient to evoke a response of the SA pathway in sweet11/sweet12 in general. Our observations suggest that the stimulation of SA accumulation can either be triggered upon challenge with Ch, or may also occur in the presence of abiotic stimuli like high humidity, as in mock treated plants.

Priming is defined as ‘a physiological state in which plants are able to faster and better activate defense responses’ (Beckers and Conrath, 2007; Moghaddam and Van den Ende, 2012). Defense priming can either be achieved by contact to avirulent pathogens or symbiotic microbes, by particular physiological conditions, or by chemical treatment (for an overview, please see Beckers and Conrath, 2007). While systemic acquired resistance (SAR) upon pathogen challenge is mediated by pipecolic acid (Bernsdorff et al., 2016), it has been shown that mobilization of SAG into free SA plays a key role in chemical priming of the SA response (Noutoshi et al., 2012). Consistently, the pool size of SAGs was 3.5-fold elevated in primed sweet11/sweet12 compared to Col-0 wild type in our study, while the difference in SAG pool size was less than 1.5-fold between double mutants and wild type in control conditions and after Ch challenge. Since a positive influence of sugar metabolism on the SA pathway is well-known (as compiled by Bolton, 2009), we assume that constantly elevated levels of soluble sugars, especially hexoses, promote defense priming in sweet11/sweet12. A stimulation of SA-dependent PR gene expression in pathogen or elicitor challenged, but not in untreated, potato tubers with antisense suppression of the plastidic ATP/ADP transporter AATP1 has previously been observed by Linke et al. (2002). Interestingly, AATP1 antisense tubers exhibited an elevated energy charge and 10-fold increased contents of glucose compared to controls (Tjaden et al., 1998). The increase in hexose content in leaves of sweet11/sweet12 double mutants was substantial throughout the diurnal cycle, but only 3.5-fold elevated compared to wild type. Nevertheless, it seems reasonable to assume a similar physiological scenario in sweet11/sweet12 double mutant leaves compared to AATP1 potato tubers. Furthermore, it has been demonstrated that priming depends on MAP kinase signaling via MPK3 and MPK6 (Beckers et al., 2009). In turn, MAPK signaling belonged to the six most significantly enriched GO terms of genes induced in mock treated sweet11/sweet12 double mutants compared to wild type, providing further evidence for priming of the SA pathway in sweet11/sweet12. The observation that the effect of the sid2 mutation on compatibility is epistatic over sweet11/sweet12 in the sweet11/sweet12/sid2 triple mutant unequivocally demonstrates that diminished susceptibility in sweet11/sweet12 is conferred by SA accumulation and the SA pathway. It is known since more than a decade that the SA triggered defense response is very effective against Ch infection during the initial biotrophic phase (Narusaka et al., 2004; O’Connell et al., 2004).