Sugar Partitioning between Ustilago maydis and Its Host Zea mays L during Infection

Ustilago maydis hijacks maize transporters to redirect plant sugars toward the infection site where it feeds, generating a sugar gradient within hyphae and causing plant yield losses.

Abstract

The basidiomycete Ustilago maydis causes smut disease in maize (Zea mays) by infecting all plant aerial tissues. The infection causes leaf chlorosis and stimulates the plant to produce nutrient-rich niches (i.e. tumors), where the fungus can proliferate and complete its life cycle. Previous studies have recorded high accumulation of soluble sugars and starch within these tumors. Using interdisciplinary approaches, we found that the sugar accumulation within tumors coincided with the differential expression of plant sugars will eventually be exported transporters and the proton/sucrose symporter Sucrose Transporter1. To accumulate plant sugars, the fungus deploys its own set of sugar transporters, generating a sugar gradient within the fungal cytosol, recorded by expressing a cytosolic glucose (Glc) Förster resonance energy transfer sensor. Our measurements indicated likely elevated Glc levels in hyphal tips during infection. Growing infected plants under dark conditions led to decreased plant sugar levels and loss of the fungal tip Glc gradient, supporting a tight link between fungal sugar acquisition and host supplies. Finally, the fungal infection causes a strong imbalance in plant sugar distribution, ultimately impacting seed set and yield.

Maize (Zea mays ssp. mays [Zm]) is a renewable source of food, feed, and industrial feedstocks domesticated from teosinte (Zea mays ssp. parviglumis; Doebley et al., 2006; Bihmidine et al., 2013). Maize yield increases reflect enhanced sugar allocation into seeds (Sosso et al., 2015), which also may have increased pathogen susceptibility during domestication (Munkacsi et al., 2008; Martinez-Soriano and Avina-Padilla, 2009). Today, the basidiomycete Ustilago maydis has both retained the ability to infect teosinte and has become a significant threat to modern maize productivity (Supplemental Fig. S1; de Lange et al., 2014). Unlike necrotrophic pathogens, which disrupt host cells to access nutrients, biotrophs such as U. maydis reprogram living cells to distribute nutrients preferentially to infection sites (Doehlemann et al., 2008a; Horst et al., 2010a, 2010b).

U. maydis infects all aerial organs, including tassels, seedlings, and adult leaves, and induces tumors that release fungal teliospores in 2 to 3 weeks. Initially, U. maydis hyphae contact plant cells using appressoria. Notably, the plant plasma membrane remains intact and enfolds the growing hyphae, generating a biotrophic interaction zone and expanding the apoplasmic area for fungal nutrient acquisition. At about 5 d post infection (dpi), host cell hypertrophy results in visible tumors (Doehlemann et al., 2008b).

In the rivalry for nutrients, pathogens and host plants have evolved opposing mechanisms. The apoplasm is a fundamental space for cell-to-cell nutrient translocation (Chen et al., 2012), and to starve pathogens, plants have evolved mechanisms to lower apoplasmic sugar concentrations (Robeson et al., 1989; Lemoine et al., 2013). For example, Arabidopsis (Arabidopsis thaliana) infected by Pseudomonas syringae, Botrytis cinerea, or powdery mildew up-regulates hexose/H⁺ symporters of the Sugar Transport Protein (STP) family (Truernit et al., 1996; Fotopoulos et al., 2003; Lemonnier et al., 2014; Yamada et al., 2016). In turn, pathogens have evolved mechanisms to increase apoplasmic sugar concentrations by exploiting host sugars will eventually be exported transporters (SWEET). Rice (Oryza sativa) OsSWEET11, OsSWEET13, and OsSWEET14 can be regulated by Xanthomonas oryzae oryzae pathovar-specific effectors, which activate OsSWEET gene transcription (Yang et al., 2006; Antony et al., 2010; Zhou et al., 2015). Also, U. maydis tumorigenesis stimulates hexose and Suc accumulation in seedling leaves (Doehlemann et al., 2008a; Horst et al., 2008, 2010a, 2010b; Matei et al., 2018). Yet, it remains unclear how sugars are allocated at the site of U. maydis infection and if the required machinery is organ specific.

The U. maydis genome encodes at least 19 predicted sugar transporters, nearly all of which are transcriptionally up-regulated during infection (Lanver et al., 2018). Two independent studies have shown that strains mutated in srt1 (sucrose transporter1) or hxt1 (hexose transporter1) genes are compromised in virulence (Kämper et al., 2006; Wahl et al., 2010; Schuler et al., 2015). The other hexose transporters did not influence virulence (Wahl et al., 2010). The high-affinity Suc transporter Srt1 is expressed exclusively when U. maydis infects its host and is able to outcompete plant sugar uptake transports. Mutation of this transporter gene does not impact U. maydis growth in axenic culture but strongly reduces tumor formation, indicating that Srt1 is required for the successful uptake of host Suc (Wahl et al., 2010). Deletion of the gene encoding the monosaccharide transporter Hxt1 also decreases virulence in planta; growth in axenic culture can be restored by expressing heterologous high-affinity Glc transporters, yet virulence is not restored, indicating that Hxt1 has additional functions as a receptor (transceptor) in planta (Schuler et al., 2015). Collectively, these studies support the hypothesis that U. maydis acquires Suc and Glc from the host via the apoplasm and that these carbohydrates are essential for fungal growth and pathogenicity.

Our goals here were (1) to study maize sugar allocation during U. maydis infection, (2) to characterize the interplay between plant and fungal sugar transporters, and (3) to monitor sugar distribution in the fungal cytosol during infection. A Förster resonance energy transfer (FRET) sensor (Fehr et al., 2003; Jones et al., 2013) has been deployed to analyze fungal Glc levels with high spatial resolution. Our data demonstrate that such sensors can be deployed in pathogens to assess metabolite dynamics in planta. To assess the impact of fungal sugar acquisition on the host, we quantified the impact of U. maydis infection on yield and found that distal infections can result in severe yield decrease.

RESULTS

U. maydis Reprograms Sugar Allocation within Maize Vegetative and Floral Organs

U. maydis infects maize vegetative and floral organs, generating large chlorotic areas and extensive tumors. Tumorigenesis is achieved by activating or hijacking host cell proliferation and expansion and by redirecting cells into a tumor pathway (Callow and Ling, 1973; Callow, 1975). In this study, we infected seedlings, expanding adult leaves, and tassels with the solopathogenic strain SG200rfp. This strain expresses a cytosolic fluorescent RFP for ease of manipulation and imaging and causes symptoms that are comparable to those seen for the wild type (Fig. 1A; Fuchs et al., 2006; Kämper et al., 2006). Seedling leaves infected at the three-leaf stage showed chlorosis, anthocyanin deposition, and tumors 14 dpi. Infection of 3- to 5-cm tassels and the surrounding whorl of not yet green leaves resulted in leaf chlorosis and tumor formation in both organs 14 dpi (Fig. 1A). Prior reports on infected seedling leaves used maize microarrays and sugar quantification assays to show that soluble sugars and starch accumulation correlated with transcriptional changes in sugar transporters. Also, those data indicate that the transition of sink to source tissue was blocked in infected seedling leaves (Doehlemann et al., 2008a; Horst et al., 2008, 2010a; Matei et al., 2018). However, numerous questions remain about the spatiotemporal sugar distribution within the tissue and the organ specificity of host responses.

Figure 1.

U. maydis affects maize organ development and sugar allocation. A, Maize vegetative and floral tissues photographed 14 dpi with U. maydis. Seedling leaves, adult leaves, and tassels are shown as indicated. B, Enzymatic soluble sugar quantification in the infected tissues (red bars) compared with mock treatment (blue bars). Values are means ± se; n = 3 technical replicates, repeated independently three times with comparable results. *, P < 0.001, by Student’s t test. FW, Fresh weight.

To learn more about organ-specific sugar accumulation, we used three different types of maize tissues: the tassel, which is a strong sink for photosynthates, as well as adult leaves and seedling leaves as classical source tissues. We classified symptoms into three categories: uninfected (no visible tumors), mild infection (sporadic conversion of flowers to tumors, up to 50% of the tassel), and severe infection (massive conversion of flowers and tassel stem tissue to tumors, 50% or more of the tassel (Supplemental Fig. S2A; Redkar and Doehlemann, 2016). For sugar quantification assays, mock-infected tassels were compared with severely infected ones. We found increased levels of hexoses and decreased amounts of Suc in infected tassels compared with mock infection (Fig. 1B). Sugar accumulation in tassels could in part explain the large weight difference observed in infected, mildly infected, and severely infected tassels: tassels exhibiting larger tumors averaged 5 times higher mass than the uninfected controls (Supplemental Fig. S2B). The same patterns of increased hexoses and decreased Suc were detected in adult and seedling leaves (Fig. 1B). Those elevated levels of tumor hexoses were in line with prior reports that examined sugar concentrations in seedlings (Doehlemann et al., 2008a; Horst et al., 2008, 2010a). In contrast to these previously published measurements on seedling leaves, no increase of Suc was detected in our study. This difference might be a result of using different maize cultivars. While previous studies used sweet corn (variety Early Golden Bantam; Doehlemann et al., 2008a; Horst et al., 2008, 2010a), we used an inbred dent variety (W23). Yet, as described previously and explained by a lack of sink-to-source transition during infection (Doehlemann et al., 2008a), the ratio of hexose to Suc was shifted toward hexoses in all three tissues.

Previous publications described elevated levels of starch accumulating exclusively at the infection site during the day (Matei et al., 2018). Under normal conditions, Calvin cycle metabolites are stored as transitory starch grains within chloroplasts during daytime, while at night, starch is nearly fully degraded and exported as Suc from photosynthetically active leaves and transported toward sink tissues. To understand if starch only accumulates during daytime in infected seedling leaves or if the diurnal pattern of starch conversion is disturbed by U. maydis infection, infected seedling leaves containing chimeric tissues with sectors containing tumors and uninfected sectors were harvested at dawn (7 am; 1 h after the end of darkness) and starch stained with Lugol’s iodine solution (iodine-potassium iodide [IKI]) or subjected to enzymatic starch quantification (Supplemental Fig. S3, A and B). By enzymatic quantification, infected leaves contained more than 4 times higher starch at dawn compared with mock-infected controls (Supplemental Fig. S3B). In contrast to uninfected leaf sections, the chloroplasts in infected areas accumulated large starch grains (Supplemental Fig. S3, C and D). We concluded that U. maydis-induced tumors contain locally copious starch possibly caused by impaired diurnal starch/soluble sugar conversion or blocked sugar leaf efflux.

U. maydis Infection Results in Expression Changes of Specific Maize Sugar Transporters

To identify the potential causes of distorted sugar allocation, we reexamined published microarray data of infected seedlings (1 and 3 dpi), adult leaves (3 and 9 dpi), and tassels (3 and 9 dpi; Skibbe et al., 2010) and selected sugar transporters belonging to well-characterized classes: SUTs (Sucrose Transporters; Riesmeier et al., 1992, 1993), STPs (Aoshima et al., 1993; Boorer et al., 1994; Toyofuku et al., 2000), and SWEETs (Chen et al., 2010, 2012). The maize genome encodes seven SUTs, 23 STPs, and 24 SWEETs (Supplemental Table S1). Thirteen of these transporters (three ZmSUTs, seven ZmSTPs, and three ZmSWEETs) showed differential mRNA levels during infection, as published previously (Skibbe et al., 2010).

ZmSUT1 is responsible for phloem loading and leaf Suc efflux in maize (Slewinski et al., 2009) and showed altered transcript levels in all three organs (tassels, leaves, and ears) at both time points tested, with significant down-regulation in seedlings and adult leaves compared with uninfected plants. STPs have already been implicated in host defense and sugar retrieval in host-pathogen interactions (Lemonnier et al., 2014; Yamada et al., 2016) and were the most affected group, with four STPs significantly up-regulated in seedlings and three down-regulated in adult leaves.

SWEETs also have been associated with plant-pathogen interactions, as it was proposed that SWEET-mediated sugar efflux is counteracted by STPs functioning as retrieval systems (Yang et al., 2006; Antony et al., 2010; Zhou et al., 2015). Three ZmSWEETs (ZmSWEET4a, ZmSWEET4b, and ZmSWEET11a) were up-regulated in infected seedlings and adult leaves and were constitutively expressed but slightly down-regulated in tassels (Skibbe et al., 2010). Based on a phylogenetic analysis of amino acid sequences, ZmSWEET4a and ZmSWEET4b belong to clade II (Eom et al., 2015), along with ZmSWEET4c, which was described previously as a plasma membrane hexose transporter (Sosso et al., 2015). ZmSWEET4a and ZmSWEET4b share greater than 80% sequence similarity with ZmSWEET4c and are likely plasma membrane hexose transporters. In contrast, ZmSWEET11a was placed by phylogenetic analysis in clade III and shares high sequence similarity (78.1%) with a known plasma membrane Suc transporter, OsSWEET11a (Yang et al., 2006). Both clade II and III members already have been implicated in supplying sugars to pathogens (Supplemental Fig. S4). Functional assays in yeast confirmed the inferences of transporter function based on phylogenetic relationships (Schuler, 2013).

To analyze ZmSWEET expression in depth, independent testing through reverse transcription quantitative PCR (RT-qPCR) assays was performed on samples from all three organ types after mock or U. maydis infection (Supplemental Fig. S5). Seedlings showed the largest ZmSWEET expression changes in response to the pathogen compared with mock infection; ZmSWEET4a and ZmSWEET4b displayed 10- and 7-fold up-regulation, respectively (Supplemental Fig. S5A). In adult leaves, the expression of ZmSWEET4a was equivalent to that in mock-infected samples, and ZmSWEET11a showed the largest change in expression (8-fold; Supplemental Fig. S5B). In tassels, modest expression changes were detected for ZmSWEET11a and significant changes for ZmSWEET4a (Supplemental Fig. S5C). These results again point to organ-specific sugar allocation mechanisms during infection.

To determine if altered ZmSWEET sugar transporter expression was restricted locally within infected organs or extended to a larger zone involving uninfected tissues, we correlated ZmSWEET gene expression levels with fungal density using RT-qPCR (Fig. 2). As a proxy for fungal biomass, RT-qPCR of the in planta-expressed U. maydis gene pit2 (protein important for tumors2) was used (Doehlemann et al., 2011). We found that samples distal to the infection site contained smaller amounts of fungal biomass compared with the proximal sample, where the expression of pit2 was 24-fold higher. pit2 expression correlated with the tumor formation pattern (Fig. 2). The expression of ZmSWEET4a and ZmSWEET4b was enriched significantly in the samples taken at the infection site compared with other sites (Fig. 2), which was consistent with the finding that starch accumulated exclusively in tumors (Supplemental Fig. S3C; Matei et al., 2018).

Figure 2.

U. maydis infection impacts the local expression of sugar transporters. Expression analysis of ZmSWEET4a and ZmSWEET4b and correlation with fungal biomass (detected by RT-qPCR using pit2 as a U. maydis reference gene) at 3 dpi are shown in three different zones (indicated with semitransparent strips) of infected and mock-treated seedling leaves. Values are means ± se; n = 3 technical replicates with expression normalized to ZmGAPDH levels, also for pit2, repeated independently three times with comparable results. *, P < 0.05, by Student’s t test.

Using a FRET Sensor to Characterize U. maydis Glc Uptake

Prior genetic analyses established that fungal Suc and Glc transporters are essential for U. maydis pathogenicity (Wahl et al., 2010; Schuler et al., 2015). Yet, sugar concentrations within fungal hyphae and the dynamics of sugar involvement in fungal growth are unknown. To study U. maydis sugar acquisition and metabolism, we deployed a Glc FRET sensor (Fig. 3A). We selected FLIPglu600μΔ13V (Chaudhuri et al., 2008) because previously it had been used successfully to report Glc accumulation in yeast and mammalian cells as well as in Arabidopsis roots. Cytosolic sugar levels were strictly dependent on three parameters: external supply, transporters, and enzymatic activities. FLIPglu600μΔ13V has a dissociation constant for Glc and Gal of ∼600 μm and can report Glc concentrations in a range between 60 μm and 6 mm. Energy transfer differences between the donor and the acceptor are detected by measuring the relative emission intensity of two fluorescent proteins (enhanced CFP and Venus) as the intensity ratio change after specific donor-only excitation (Dx). FLIPglu600μΔ13V displays a negative ratiometric change after Glc addition. The U. maydis strain SG200 was stably transformed with FLIPglu600µΔ13V under the control of the strong constitutive Hxt1 promoter by integration into the ip-locus (the resulting strain was named SG200P_hxt1:FLIPglu600µΔ13V, hereafter SG200glu600).

Figure 3.

Analysis of FRET Glc sensor U. maydis strain SG200glu600 in axenic culture. A, Cartoon of Glc FRET sensor FLIPglu600μΔ13V before and after Glc addition. CFP (donor, D) and Venus (acceptor, A) are drawn as blue and yellow barrels, while the Glc-binding domain of the sensor (mature MglB from Escherichia coli) is displayed in red. B, SG200glu600 sporidia were imaged in axenic culture using three channels: AxAm, Venus was excited at 505 nm and emission was detected at 525 nm; DxDm, CFP was excited at 428 nm with CFP emission recorded at 480 nm; DxAm, CFP was excited at 428 nm and acceptor Venus emission was recorded at 525 nm. Bars = 10 μm. C, Axenic cultures of SG200glu600 were perfused with 10 mm Suc, Glc, or Fru. Calculated DxAm/DxDm showed a significant decrease in ratiometric value only after Glc addition (n = 12). *, P < 0.05, by Student’s t test. D, Glc was added at three different concentrations (1, 10, and 100 mm) to SG200glu600 in axenic culture, and FLIPglu600μΔ13V DxAm/DxDm was measured (n = 9). *, P < 0.05, by Student’s t test. E, Cells were grown as described for C and D, but instead of applying different sugars or different Glc concentrations, we changed the pH of the medium. None of the different pH treatments triggered a significant ratiometric response (n = 8).

To validate sensor expression and function, SG200glu600 was tested in axenic culture. By confocal microscopic imaging, uniform Venus (acceptor excitation-acceptor emission [AxAm]) fluorescence was observed in the cytosol of most fungal cells, with no detectable vacuolar accumulation (Fig. 3B). For functional analysis, fluorescence emissions of SG200glu600 and the progenitor strain SG200 were measured in plate assays after the addition of different monosaccharides or disaccharides or water. SG200 background fluorescence values were subtracted from the measured SG200glu600 values, and DxAm/Dx donor emission (Dm) ratios were calculated. The addition of Glc resulted in a significant decrease in the DxAm/DxDm ratio, while the addition of Suc and Fru had no significant effects (Fig. 3C). We also performed a titration analysis using four different Glc concentrations from 0 to 100 mm (Fig. 3D). A significantly different DxAm/DxDm ratio was measured after adding 1 and 10 mm Glc, while no statistically significant ratio differences were observed between 10 and 100 mm Glc. Under four different pH conditions, ranging from pH 5 to pH 8, no significant differences of sensor performance were detected (Fig. 3E). Disease ratings in planta demonstrated that SG200glu600 induced tumors at rates comparable to SG200rfp (Supplemental Fig. S6). Collectively, these experiments indicate that the expression of FLIPglu600µΔ13V in U. maydis does not impact virulence and that ratiometric changes in sensor fluorescence are Glc dependent in axenic cultures.

The U. maydis Cytoplasmic Glc Gradient Is Related to Plant Glc Levels

To demonstrate sensor function in planta, seedling leaves were inoculated with strain SG200glu600. In planta, older U. maydis hyphae parts are partitioned from actively growing regions by septation, and with time, older hyphae segments lose biological activity. We performed in vivo imaging on septa-forming hyphae (Supplemental Fig. S7). Each hyphal image was collected using three different channels: AxAm (excitation, 500 nm; emission, 525 nm), DxDm (excitation, 428 nm; emission, 480 nm), and DxAm (excitation, 428 nm; emission, 525 nm). To evaluate the FRET sensor response, we generated ratiometric images: intensity values from the DxAm channel (exciting the donor CFP, recording the acceptor Venus emission) were divided by values from the DxDm channel (exciting the donor CFP, recording the donor CFP emission). The images generated were colored with a 16-color palette, representing ratiometric values (Figs. 4 and 5; Supplemental Fig. S7). Clear differences in ratiometric values were detected in the different segments, with the highest ratio in the oldest hyphal fragment or clamps and the lowest ratios detected in the hyphal tip (Supplemental Fig. S7). These measurements indicate that the oldest hyphal fragment or clamps should contain less Glc than the growing tip region, as we would expect with the tip being more biologically active.

Figure 4.

Ratiometric readout of a FRET Glc sensor expressed in U. maydis cytosol. A, In planta confocal imaging of U. maydis strain SG200glu600 hyphae infecting seedlings at 3 dpi, grown under 14 h of light/10 h of dark. The DxDm and DxAm channels were used to derive the ratio image (DxAm/DxDm) describing three distinct hyphal zones. Hyphal base, center, and tip are labeled in blue, yellow, and red, respectively. Bars = 20 μm. B, Ratiometric measurements were performed for all hyphae present in A. Histograms show the ratio quantification of the first (base), central (center), and last (tip) 5 μm of each hypha. Values are means ± se; n = 18. *, P < 0.05, by Student’s t test. C, Location of the lowest DxAm/DxDm value across 75 different hyphae, reported in percentage from measurements of infected, light-grown seedlings.

Figure 5.

Light conditions affect carbohydrate accumulation in maize and fluorescence ratios as measured using a FRET Glc sensor expressed in U. maydis. A, Diagram of the experimental protocol and light regimes used. B, Quantification of leaf carbohydrates after treatment as described in A. Values are means ± se; n = 3 technical replicates, repeated independently three times with comparable results. FW, Fresh weight; nd, not detected. C, For each of the three treatments, hyphae were imaged and DxAm/DxDm ratios were calculated in three hyphal areas (base, center, and tip, 5 μm each). *, P < 0.05, by Student’s t test. D, Distribution of the lowest DxAm/DxDm value as a proxy for Glc maxima in three U. maydis populations grown under previously described conditions (n = 75, 33, and 20).

To analyze Glc distribution in U. maydis hyphae during infection in detail, we performed 16 independent infection and imaging trials (at 3 dpi), using different batches of maize seedlings, generating images for 128 unique hyphae (Fig. 4). Three days postinfection, U. maydis should have thoroughly colonized epidermal cells and initiated mesophyll cell infection. For analysis, we only selected hyphae that (1) had penetrated epidermis or mesophyll cells, (2) showed homogenous sensor distribution, and (3) the direction of hyphal growth could be identified. We defined three 5-μm-long regions: base, center, and tip. The tip region was defined as the first, youngest 5 μm of the hypha. The center represents 5 μm around the arithmetical middle of the hypha in 2D projection, and the base was defined as the last, oldest 5 μm of hypha where fluorescence in all channels could be detected (Fig. 4A). Eighteen counts and ratio determinations were performed per region. The majority of hyphae (65.4%) showed a significantly lower ratio at the tip compared with the center and base (Fig. 4, B and C), indicative of a higher Glc concentration. There were nonsignificant differences among the sections in 24% of cases, while 9.33% and 1.33% had the lowest ratio point at the base and the center, respectively. These results indicate that Glc accumulates in the growing tip region of the hyphae and are in line with our previous data showing higher Glc concentration at the fungal growing tip rather than the older base clamps.

To test if the recorded differences in Glc distribution were linked to fungal development and its intimate relationship with plant metabolic status, we perturbed plant sugar metabolism with a dark treatment to halt photosynthesis and block diurnal sugar production and storage. Nine-day-old seedlings were infected with SG200glu600, left 1 d under standard light conditions, and then maintained in this condition (75 seedlings shown in Figs. 4C and 5D) or transferred to darkness for 2 d. In contrast to infected plants grown under standard light conditions, no soluble sugars were detectable in dark-grown seedlings (Fig. 5B). In dark-treated seedlings, the majority of hyphae (48.5%) had nonsignificant ratio differences between base, center, and tip (Fig. 5, C and D). To establish if the differences in DxAm/DxDm ratio distribution between light- and dark-treated plants were Glc dependent or resulted from other changes during the dark treatment, infected dark-treated seedling sections were incubated in 10 mm Glc for 20 min prior to imaging (Fig. 5A). This sugar treatment replenished leaf Glc concentration to a level 2-fold higher than in light-grown seedlings (Fig. 5B). We imaged 20 hyphae after sugar replenishment and compared the average ratio for each region with the previously determined values for light- or dark-treated seedling hyphae. On average, these hyphae showed the lowest DxAm/DxDm ratio at the tip. Overall, light-grown and dark-grown plus 10 mm Glc treatment resulted in significant sensor fluorescence ratio differences between the three defined regions compared with the dark-grown treatment (Fig. 5C). The tip ratio fluctuated significantly depending on plant Glc levels. It is noteworthy that, in all treatments, we also recorded 9.3%, 18.2%, and 5% of hyphae in which the ratio was significantly higher in the older parts; further experimentation will be necessary to understand the nature of this signal (Fig. 5D).

In summary, our findings indicated that the reported differences in DxAm/DxDm ratios along the hypha, and in response to different treatments, correlated with Glc concentration in the hypha. Under normal conditions, Glc concentrations were highest at the fungal tip, while under Glc-limiting conditions, Glc appeared to be more homogenously distributed.

U. maydis Distal Tassel Infection Impacts Seed Set and Yield

After describing how sugar distribution changed following plant metabolic changes, we were interested to determine if the fungal sugar diversion would have an impact on plant sugar allocation and yield. To our knowledge, the impact of U. maydis distal infections on yield in individual plants had not been quantified. Thus, we performed field trials to measure maize kernel weight and number after tassel infection by U. maydis strain SG200rfp, and we compared these data with detasseled plants, used here as a control (Supplemental Fig. S8A). By detasseling, we removed a strong sink tissue that competes with the ear for plant metabolites, and this manipulation has been speculated to boost yield. In three biological replicates over two field seasons (two in 2014, fields a and b spatially and temporally separated, and one in 2015), significant yield losses were observed in the infected groups compared with controls. Uninfected yield ranged from 11.6 g per plant in 2014a to 21.4 g per plant in 2015. Infected plants showed yield reductions between 30% in 2014a and 47% in 2015 (Supplemental Fig. S9).

To determine whether yield changes reflected single kernel weight or number of kernels per ear, we determined how many plants had fully developed ears and how many kernels per ear were present for each trial (Supplemental Figs. S8B and S9). Strikingly, in all three plots, the proportion of plants with poor or no yield (fewer than 20 seeds per ear) was highest in the U. maydis-infected samples. Within the uninfected and detasseled populations, only 3% to 10% of plants had poor yield. In contrast, in infected samples, 45.1% (2014a trial; Supplemental Fig. S8B), 39.2% (2014b), and 90% (2015) exhibited poor yield. Unusually warm nights after fungal infection in 2015 may have resulted in more deleterious yield losses. In summary, we found that U. maydis infections in tassels resulted in substantially reduced yield primarily by impacting the number of kernels per ear, likely by diverting metabolites from the developing ear toward infected tassels.

DISCUSSION

The biotrophic fungus U. maydis promotes massive host cell proliferation and cell expansion, resulting in fungus-induced tumors. The goal of the fungus is to feed on plant resources, especially sugars and free amino acids (Doehlemann et al., 2008a; Horst et al., 2010a; Skibbe et al., 2010). These observations lead to the following questions: How does the fungus acquire these nutrients and how are they metabolized? Does metabolite redirection significantly impact the host?

In seedling leaves, we found tumors to be largely filled with soluble sugars and starch, with starch clearly colocalizing with tumors (Fig. 1; Supplemental Fig. S3). Our cytological analysis found extensive cell hypertrophy and persistent starch granules surrounding infected leaf vasculature (Supplemental Fig. S3). A recent more detailed cytological analysis (Matei et al., 2018) quantified both hypertrophy and an increase in the number of some leaf cell types. In line with this, a previous publication reported increased expression of genes encoding starch-producing enzymes in U. maydis-infected seedlings (Doehlemann et al., 2008a; Kretschmer et al., 2017). Both high soluble sugar concentration and persistent starch have been linked to defective leaf sugar efflux, as discovered in maize and Arabidopsis mutants lacking vasculature-specific sugar transporters (Slewinski et al., 2009; Chen et al., 2012). Specifically, maize ZmSUT1 was shown to be responsible for at least one step in phloem loading. Zmsut1 loss-of-function mutants are defective in Suc leaf efflux, resulting in persistent starch and high soluble sugars in leaves (Slewinski et al., 2009), similar to U. maydis-infected leaves. ZmSUT1 is significantly transcriptionally reduced in infected leaf tissues, possibly impairing sugar efflux from infected leaves and generating the persistent pool of surplus sugars and starch we measured.

In contrast to observations made in seedlings here and previously (Doehlemann et al., 2008a; Horst et al., 2010a), we found reduced levels of Suc in infected sink tissues (tassels). In line with this, ZmSUT1 expression is not changed significantly at 3 dpi in the tassel. These data suggest that the organ specificity of host responses to U. maydis infection is not limited to target defenses (Skibbe et al., 2010) but also includes metabolic pathways.

The fungus must access sugars sequestered in host cells, and this step likely involves the reprogramming of host sugar transporters: seven ZmSTPs and three ZmSWEETs are differentially regulated in infected tissues. These two classes of transporters have already been linked to pathogenesis. Because they are H⁺/monosaccharide transporters, STPs have been speculated to boost hexose retrieval from the apoplasm. This, combined with increased plant invertase activity, could limit sugar availability to pathogens (Truernit et al., 1996; Fotopoulos et al., 2003). In contrast, SWEET transporters have been positively linked to pathogen nourishment (Chong et al., 2014; Cohn et al., 2014; Hu et al., 2014) as well as in feeding beneficial rhizobia or arbuscular mycorrhizal symbionts in Medicago truncatula (Kryvoruchko et al., 2016) or Solanum tuberosum (Manck-Götzenberger and Requena, 2016), respectively. Three maize SWEET transporters were up-regulated during seedling infections (Supplemental Fig. S5A), likely increasing both Suc and hexose allocation. ZmSWEETs were locally up-regulated only in proximity to infection zones (Fig. 2). In tassels, only ZmSWEET4a expression was increased, once more reflecting organ-specific disease strategies. Further studies are now required to dissect this relationship, using sugar transporter-reporter lines or detailed in situ RNA hybridization and immunolocalization during an infection time course, to determine precisely how closely the transporter overexpression localizes at the biotrophic interface. Ultimately, the use of knockout lines could explore to what extent U. maydis depends on each of the sugar transporters, especially the three ZmSWEETs investigated here.

What type of fungal signal promotes the plant to transcriptionally deregulate sugar transporters? In rice, orange (Citrus spp.), and cassava (Manihot esculenta), bacteria of the genus Xanthomonas (X. oryzae, X. citri, and X. axonopodis) hijack SWEET gene transcription by deploying type III Transcription Activator-Like effectors, which bind the promoter of target SWEETs and trigger ectopic overexpression. The U. maydis genome lacks obvious Transcription Activator-Like effector-like genes (Kämper et al., 2006). Therefore, future studies will be necessary to identify if the transcription of ZmSWEETs, as well as other maize sugar transporters, is regulated by U. maydis effectors directly or reprogrammed by plant signaling after infection. Hijacking sugar transporters is just the first step in generating stable access to plant carbohydrates, and subsequent steps involve sugar uptake from the plant apoplasm into the fungal cytosol via fungus-encoded sugar transporters (Wahl et al., 2010).

To explore potential links between sugar metabolism, fungal pathogenicity, and development, we deployed a cytoplasmic Glc biosensor in U. maydis. Stable transformation within the fungal ip-locus resulted in strong and homogenous sensor distribution throughout the fungus (Figs. 3B and 4A, AxAm), allowing us to monitor cytoplasmic Glc levels within living hyphae. In axenic culture, the sensor-expressing U. maydis strain (SG200glu600) showed decreasing fluorescence ratios (DxAm/DxDm) after the addition of externally supplied Glc in a concentration-dependent manner (Fig. 3D). This effect was independent of pH changes (Fig. 3E). While imaging hyphae growing within the plant, we detected a clear trend of sensor ratiometric response, with most hyphae showing the lowest ratio at the young hyphal tip and in newly septated branches, compared with older parts of the same hypha (Figs. 4 and 5; Supplemental Fig. S7). In hyphae infecting dark-grown plants (with no detectable leaf sugar levels), we found that most showed nonsignificant ratio differences among base, center, and tip. By applying Glc to sections of dark-grown leaves, we monitored the reestablishment of the ratio gradient (lower at the tip, higher at the base and center) similar to U. maydis-infected light-grown seedlings full of soluble sugars. The association between Glc presence and the sensor ratiometric response suggested that the DxAm/DxDm changes can be used to measure differences in Glc distribution within the U. maydis hypha. Yet, we cannot entirely rule out that ratio differences could be influenced by optical artifacts or changing sensor properties (e.g. pH or ionic changes).

Based on current observations, we propose that the lower fluorescence ratios within hyphal tips reflect elevated levels of Glc compared with other regions. Several mechanisms that result in a Glc gradient can be hypothesized: (1) the fungal sugar import machinery could be tip polarized, (2) vesicular traffic may carry and recycle more sugar transporters at the tip, or (3) sugar metabolism rates could differ. In support of the first possibility, U. maydis Hxt1 and Srt1 have been localized at the plasma membrane, but the reporter lines used do not have the resolution to discriminate if there is transporter polarization. In the fava bean (Vicia faba) rust Uromyces fabae, the hexose transporter Hxt1 has been described to be tip polarized (Voegele et al., 2001). Further investigations that go beyond the scope of this study are needed to understand transporter localization during infections. With regard to vesicular trafficking, U. maydis growth is linked to massive vesicular trafficking at the hyphal tip. It was demonstrated that high rates of endocytosis and exocytosis determine the polarization of fungal effector release from the hyphal tip (Treitschke et al., 2010). In addition, in other rapidly growing structures such as plant pollen tubes (Derksen et al., 1995) and root hairs (Ketelaar, 2002), polarized vesicle traffic is involved in local protein accumulation. Thus, it is possible that high vesicle trafficking rates could be responsible for carrying and recycling more transporters.

In considering the possibility of differential Glc utilization within hyphae, we note that high vesicle numbers in growing pollen tubes as well as in elongating hyphae result in the exclusion of most organelles such as mitochondria from the apex (Howard, 1981; Selinski and Scheibe, 2014). The exclusion of mitochondria may result in alternative energy generation and consumption at the tip. Localization and counting of both mitochondria and Hxt1 proteins (using transmission electron microscopy and gold immunolabeling of Hxt1 proteins) within the hyphal tip could help clarify this hypothesis.

In Arabidopsis, tubular root cells extend by tip growth and play an important role in sensing extracellular biotic and abiotic conditions, including nutritional status (Shin et al., 2005). Polarized growth is dependent on gradients as a result of tip-localized uptake (Wymer et al., 1997). Taking this into account, we speculate that the Glc gradient within the hyphae could be important for polar growth. The bias within the hyphae may direct fungal growth toward plant tissues containing more sugars (e.g. veins). This idea is in line with the observation that fewer hyphae with a tip-bottom Glc gradient were present in dark-treated plants.

In conclusion, we demonstrated that U. maydis has a strong negative impact on maize sugar physiology and yield, diverting sugars from the developing ears to infected organs where a sugar-rich niche is associated with growing fungi. Based on our results, we propose the following model for U. maydis and maize rivalry for carbon. U. maydis triggers local up-regulation of SWEET transcription; SWEET transporters are then recruited to leak sugars into the apoplasm at the biotrophic interface. Simultaneously, ZmSUT1 is repressed, (1) preventing immediate Suc retrieval and (2) impairing sugar loading into the phloem, thereby allowing apoplasmic sugar buildup. From this sugar-rich apoplasm, U. maydis utilizes its sugar transporters to acquire soluble sugars, concentrating at least Glc at the growing hyphal tip. In plants with blocked Glc production, most fungal hyphae analyzed lacked differences in cytosolic Glc distribution, providing further evidence of the tight association between plant and fungal metabolism in the rivalry for carbon.

MATERIALS AND METHODS

Plant Cultivation and Treatment with Glc

The inbred maize (Zea mays) line W23 bz2 (defective in vacuolar anthocyanin sequestration) is maintained by self-pollination. Seedlings were grown under Stanford University greenhouse conditions: 14-h day, 10-h night, 28°C day, and 22°C night, with lighting as described (Casati and Walbot, 2003). Adult plants were analyzed under field conditions at Stanford in summers 2014 and 2015. Experiments 2014a and 2014b were performed in temporally separated plots in the same field. For dark treatment experiments, 1-dpi seedlings were transferred to complete darkness for 2 d at the above-described temperature conditions. Corresponding control plants remained under the above-mentioned light cycle and temperature conditions. Following the light or dark treatment, 2 cm of Ustilago maydis-infected seedling tissue was excised 1 cm below the infection site and submerged in 10 mm Glc solution for 20 min. After Glc incubation, leaf samples were washed individually five times in water and excess water was wiped off. Samples were frozen in liquid N₂ or directly subjected to microscopic imaging.

Generation of U. maydis Strain SG200glu600

FLIPglu600µΔ13V (Chaudhuri et al., 2008) was amplified by PCR with primers Glucμ600_attB1_for and NanoS_atttB2_rev (Supplemental Table S2) and cloned into pDONOR221 (Invitrogen). The resulting vector was used together with pH-Dest to generate expression vector pEXPR-glu600 by LR cloning (Invitrogen). The final expression vector was stably integrated in the ip-locus of U. maydis strain SG200 (Kämper et al., 2006), resulting in strain SG200glu600. Correct integration of the construct was verified by Southern blotting, showing multiple integration of the construct to express the Glc sensor FLIPglu600µΔ13V under the control of the hxt1 (Schuler, 2013; Schuler et al., 2015) promoter.

U. maydis Cultivation and Infections

In this study, U. maydis strains SG200glu600, SG200 (Kämper et al., 2006), and SG200rfp (Fuchs et al., 2006) were used. SG200rfp was used in most experiments as a control and shows a wild-type-like infection pattern; by fluorescence imaging, SG200rfp is readily distinguished from wild, natural infections in the field. For seedling infections, liquid cultures of strain SG200glu600, SG200, or SG200rfp were grown in YEPS_light (Tsukuda et al., 1988), shaking at 200 rpm at 28°C, to an optical density (OD)_600nm of 0.8 to 1. Cells were sedimented at 2,500g for 5 min and resuspended in water to an OD_600nm of 1, when used for disease ratings (Kämper et al., 2006), or to OD_600nm of 3, if used for confocal imaging, sugar measurement, or starch staining. Seedling infections were performed 9 d after sowing. Water was used for mock infection controls. Uninfected refers to plants that received no treatment. Field tassel infections, which inevitably also cause infections in still expanding adult leaves, followed (Redkar et al., 2015), with strain SG200rfp (Fuchs et al., 2006) resuspended in water to an OD_600nm of 1.

Yield Analyses

The W23 bz2 inbred requires about 60 d from seed sowing to pollen shed. For each field trial, 120 W23 seeds were field sown or planted in 20-cm containers for transplantation on days 12 to 14 postsowing. Groups of 90 plants matched in size were selected and divided into three treatment groups. Control plants had no treatment; detasseled plants had the tassel removed manually when it was approximately 6 cm above the leaf canopy, about 10 to 14 d prior to pollen shed by control plants. For the U. maydis-infected group, a fungal suspension was injected into the air space surrounding 3- to 5-cm tassels (approximately day 35); at this stage, the tassel contains anthers of 300 to 600 µm; pollen shed would normally start about 25 d later. The ears were hand pollinated on 2 successive days with copious W23 bz2 pollen. Seeds were left to dry on the ear for about 30 d after pollination, then harvested ears were dried for about 2 weeks at 40°C. Seeds from each ear were harvested, counted, and weighed to perform yield evaluation.

Sugar Quantification

Samples of different tissue types after U. maydis or mock infection (organ and infection conditions indicated in the corresponding figures) were frozen immediately in liquid nitrogen after harvest, ground to a fine powder with mortar and pestle in liquid nitrogen, and 50- to 60-mg aliquots were incubated with 500 µL of 70% (v/v) ethanol for 1 h on ice, with frequent and vigorous vortexing. Subsequently, the samples were centrifuged for 10 min at 13,000g, and the supernatant was recovered. This process was repeated once, then supernatants were dried in a vacuum concentrator and resuspended in 500 µL of water. Sugars were quantified with the Suc/d-Glc/d-Fru Assay Kit protocol, UV method (Roche).

Starch Staining

Infected and mock-infected seedlings were harvested at 7 am (dawn). Freshly cut leaves were boiled in 95% ethanol for approximately 30 min, until chlorophyll pigments disappeared. Cleared leaves were submerged in saturated IKI for 15 min, then rinsed twice with water, and imaged with a Lumix GF1 camera (Panasonci). The IKI used for starch staining was made by adding 1 g of iodine and 1 g of potassium iodide to 100 mL of water.

Plastic Embedding and Sectioning

Maize leaves (infected with SG200rfp) collected at 7 am were placed into a fixative solution of 0.1 m cacodylate buffer with 2% (v/v) paraformaldehyde and 2% (v/v) glutaraldehyde, vacuum infiltrated for 15 min, and left in the fixative solution overnight. Dehydration followed the fixation step, with an increasing ethanol gradient (10%, 30%, 50%, 75%, and 95%). Plastic embedding was performed according to the LR White embedding kit protocol (Electron Microscopy Science). Semithin transverse sections (1.5 μm) were cut using Ultracut (Reichert [now Leica]), stained with 0.1% Safranin O for 30 s, and then washed twice with distilled water. All sections were mounted with CytoSeal 60 (Electron Microscopy Science).

Phylogenetic Analyses

Protein sequences were retrieved from National Center for Biotechnology Information Blast (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and Phytozome (https://phytozome.jgi.doe.gov/pz/portal.html) using AtSWEET11 as the search sequence, and their accession numbers are listed in Supplemental Table S1. The evolutionary history was inferred using the neighbor-joining method. The optimal tree with the sum of branch length = 14.56385436 is shown. The evolutionary distances were computed using the JTT matrix-based method (Jones et al., 1992) and are in units of the number of amino acid substitutions per site. The analysis involved 67 amino acid sequences. All positions with less than 95% site coverage were eliminated. That is, fewer than 5% alignment gaps, missing data, and ambiguous bases were allowed at any position. There were 213 positions in the final data set. Evolutionary analyses were conducted in MEGA6 (Tamura et al., 2013).

RT-qPCR Analysis of Candidate Gene Expression

For the initial analysis of ZmSWEET expression 3 dpi in seedlings, adult leaves, and tassels, a First-Strand cDNA Synthesis Kit (Roche; Skibbe et al., 2010) was used with primers Sweet4a_RT_f, Sweet4b_RT-r, Sweet4b_RT_f, Sweet4b_RT_r, Sweet11_RT_f, Sweet11_RT_r, cyanase_RT_f, and cyanase_RT_r (reference gene). RT-qPCR was performed in two biological replicates with two technical replicates using the IQ SYBR Green Supermix (Bio-Rad). Cycling conditions were as follows: 7 min at 95°C, followed by 45 cycles of 30 s at 95°C, 20 s at 60°C, and 40 s at 72°C. Gene expression levels were calculated using PCR Miner Software (Zhao and Fernald, 2005) using the ZmCyanase gene as the reference. For local seedling leaf-specific expression of ZmSWEETs, 2-cm-long leaf samples were retrieved from three locations: 3 cm above the injection site and 1 and 9 cm below the infection site. Leaf samples treated with SG200rfp or mock (water) were harvested at 11 am 3 dpi. Total RNA was extracted using the Spectrum Plant Total RNA kit (Sigma). First-strand complementary DNA was synthesized using the QuantiTect Reverse Transcription kit following the instructions of the supplier (Qiagen). The same primers as above were used for RT-qPCR to determine expression levels using the LightCycler 480 System (Roche). Primers specific for pit2 were used to assess fungal presence in the same samples, using ZmGAPDH as the reference gene. The 2^−ΔΔCt method was used for relative quantification (Pfaffl, 2001). All primers are listed in Supplemental Table S2.

Fluorimetric Analysis of the Growth of U. maydis Strains Expressing FRET Sensors

Logarithmically growing cultures of strains SG200rfp and SG200glu600 were harvested by centrifugation at 3,500g for 10 min, washed three times in water, and resuspended in water to a final OD_600nm of 5. After starvation for 1 h (200 rpm, 28°C), 135 µL of U. maydis cell suspension was transferred onto a 96-well microplate for analysis in a Tecan Infinite M200 PRO plate reader (Tecan); water, Suc, Fru, or different concentrations of Glc were added, resulting in a final volume of 150 µL. After 1 min, cell density was measured at 600 nm. Emission spectra were recorded with CFP excitation (428 nm) between 470 and 600 nm and with Venus excitation (500 nm) between 520 and 600 nm. Single emission measurements were performed at 480 nm (Dm) and 525 nm (Am) with 428 nm excitation (Dx) and at 525 nm with 500 nm excitation (AxAm). All measurements were performed with 10 flashes and 20-μs integration time in at least six technical repetitions with two biological replicates. Background was calculated by subtracting the fluorescence intensity measured for SG200rfp from the intensity measured for SG200glu600. DxAm values were divided by DxDm to generate a ratiometric value to plot.

Fluorescence Imaging, Image Processing, and Analyses

For ratio imaging, infected leaf areas of seedlings 3 dpi with SG200glu600 were dissected. Imaging was performed on an SP8 with an AOBS system with HyD SMD detectors using a 63× water objective (Leica Biosystems) in 0.5-µm z-stacks. The CFP (donor) was excited at 428 nm, with donor emission recorded from 450 to 480 nm and acceptor (Venus) emission recorded from 525 to 575 nm. To analyze if there was an even distribution of FLIPglu600µΔ13V in SG200glu600 hyphae, Venus was excited at 500 nm and emission was detected between 525 and 575 nm. Image processing and analysis were performed using FIJI (http://fiji.sc/). Mean intensity values for regions of interest were calculated as follows. The z-stacks of the DxDm and DxAm channels were summed before taking the ratio of the summed DxAm and summed DxDm. A mask generated using the AxAm channel (Phansankar filter) was applied to the ratiometric images. Average ratios of DxAm/DxDm from the tip versus older part of the hyphae were calculated for each image, tracing three- to five-pixel wide segments along the hyphae center, to create a plot with ratio values along the whole hyphae length. Ratio mean and sd were calculated for the first, central, and last 5 μm of each hyphae.

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL data libraries under the accession numbers listed in Supplemental Table S1.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. U. maydis-infected teosinte tassel.

• Supplemental Figure S2. U. maydis infections affect tassel morphology and density.

• Supplemental Figure S3. U. maydis impairs starch breakdown and induces cell proliferation and hypertrophy.

• Supplemental Figure S4. Phylogenic analysis of maize, rice, and Arabidopsis SWEETs, and orthologs involved in plant-pathogen interaction.

• Supplemental Figure S5. Expression analyses of three ZmSWEET transporters induced during U. maydis infection.

• Supplemental Figure S6. Disease ratings of infections with SG200rfp and SG200glu600.

• Supplemental Figure S7. U. maydis septation pattern coincides with step changes in ratiometric readout.

• Supplemental Figure S8. U. maydis-infected plants show reduced productivity.

• Supplemental Figure S9. Additional yield data from field trials 2014b and 2015.

• Supplemental Table S1. Accession numbers.

• Supplemental Table S2. Primer sequences.