PsANT, the adenine nucleotide translocase of Puccinia striiformis, promotes cell death and fungal growth

Adenine nucleotide translocase (ANT) is a constitutive mitochondrial component that is involved in ADP/ATP exchange and mitochondrion-mediated apoptosis in yeast and mammals. However, little is known about the function of ANT in pathogenic fungi. In this study, we identified an ANT gene of Puccinia striiformis f. sp. tritici (Pst), designated PsANT. The PsANT protein contains three typical conserved mitochondrion-carrier-protein (mito-carr) domains and shares more than 70% identity with its orthologs from other fungi, suggesting that ANT is conserved in fungi. Immuno-cytochemical localization confirmed the mitochondrial localization of PsANT in normal Pst hyphal cells or collapsed cells. Over-expression of PsANT indicated that PsANT promotes cell death in tobacco, wheat and fission yeast cells. Further study showed that the three mito-carr domains are all needed to induce cell death. qRT-PCR analyses revealed an in-planta induced expression of PsANT during infection. Knockdown of PsANT using a host-induced gene silencing system (HIGS) attenuated the growth and development of virulent Pst at the early infection stage but not enough to alter its pathogenicity. These results provide new insight into the function of PsANT in fungal cell death and growth and might be useful in the search for and design of novel disease control strategies.

Scientific RepoRts | 5:11241 | DOi: 10.1038/srep11241 and cytosolic (c) states, in which ADP-ATP is bound to either the matrix side or the cytosolic side of the inner membrane, respectively. The MPT activity is determined by the orientation of the nucleotide site of ANT. Thus, carboxyatractyloside favors the c-state of ANT-induced MPT, whereas bongkrekate favors the m-state of ANT-inhibited MPT 8 .
The role of ANT in the regulation of cell death by MPT is conserved across nematodes and mammals. In Caenorhabditis elegans, there are four isoforms of worm ANT (WAN-1, 2, 3 and 4), but only the knockout of WAN-1 yields a lethal phenotype. Shen et al. 9 reported that WAN-1 forms a complex with CED-4 (the nematode counterpart of Apaf-1) and CED-9 (a Bcl-2 ortholog). The BH3-only protein EGL-1 disrupts the interaction between CED-9 and WAN-1, which leads to osmotic swelling of the mitochondrial matrix and eventually to the permeabilization of mitochondrial outer membrane. Four closely related isoforms of ANT (ANT1, 2, 3 and 4) exist in humans, and ANT1 and ANT3 are pro-apoptotic proteins 10,11 , whereas ANT2 exerts pro-survival functions 12 . The ability of ANT4 to modulate apoptosis has not yet been investigated. The main processes and key players that drive programmed cell death and apoptosis in mammalian cells display a high level of conservation throughout evolution. However, some important differences exist, such as the release of cytochrome c (Cyt c) from the mitochondrial intermembrane space, an event that seems to occur exclusively in mammalian cells 5 . Meanwhile, genetic evidence has suggested that ANT makes a phylogenetically old contribution to the cell death machinery in Saccharomyces cerevisiae. The absence of all three ANT isoforms (aac1, aac2 and aac3) in yeast protects cells exposed to acetic acid and diamide from cell death 13 .
Functional analyses of genes in yeasts revealed that the progression of apoptotic -like programmed cell death (PCD) in the single-celled fungus yeast resembles that of higher eukaryotes at the molecular level 14,15 . Apoptotic cell death also occurs in the model pathogenic fungi Candida albicans 16,17 , Aspergillus fumigates 18,19 and Magnaporthe oryzea 20 . The discovery of the apoptosis response makes it possible to control fungal diseases by developing novel antifungal drugs and fungicides that could activate fungal cell apoptosis. However, in contrast to the increasing number of studies that are beginning to unravel the apoptosis pathway in yeasts, comparatively little work has focused on pathogenic fungi. Comparative analysis of the genome sequence of four Aspergillus and seven other fungal species revealed more than 100 apoptosis-associated genes, including lineage-specific proteins as well as a conserved core component of the ancestral apoptosis machinery that is shared by all fungi, suggesting that there is a complex uncharacterized regulatory network in fungi 21 . In addition, homologs of several mammalian apoptotic proteins, including PARP and AMID, have been found in filamentous fungi 21 , indicating that the molecular pathways that control apoptosis in different organisms are conserved even though they differ in complexity. Furthermore, the overexpression of Cyclophilin D PaCypD accelerated organism aging and cell death induction in Podospora anserina 22 . The deletion of PaAif2 and PaAmid2, two genes encoding mitochondrial apoptosis-inducing factors (AIFs) such as oxidoreductases, led to increased stress tolerance and lifespan extension 23 . In addition, bioinformatic screening of fungal genomes also revealed that many of the known components of apoptosis in higher eukaryotes are missing or show high divergence, indicating that pathogenic fungi may possess a distinctive apoptosis pathway.
Identification of the molecular switches that trigger fungal apoptosis is of paramount importance in generating drug therapies. However, despite the existence and cell-death-regulating function of CYPD and AIF in fungi, little is known about the cell death machinery in pathogenic fungi. In this study, we obtained an ANT homolog PsANT in the pathogenic rust fungi Puccinia striiformis f. sp tritici (Pst) which devastates wheat crops worldwide 24 . The PsANT protein contains three typical mito-carrier domains and is located in the mitochondria of infected Pst hyphae. PsANT exhibited a pro-cell-death effect in fission yeast, wheat and tobacco cells. Furthermore, knockdown of PsANT in Pst using host-induced gene silencing (HIGS) blocked the growth and development of Pst in the early infection stage but did not alter the virulence of Pst. A multi-sequence alignment with various ANT sequences in the National Center of Biotechnology Information database and Broad Institute database revealed that PsANT shared more than 70% identity with its orthologs from other fungi, including Mlp, Puccinia graminis f. sp. tritici (Pgt) and Saccharomyces cerevisiae (Fig. 1a). In contrast, PsANT had only 51.68% identity with HsANT1 in humans and 46.4% to WAN2 in C. elegans. The primary structure of PsANT contained three conserved 100-amino-acid-long mito-carr domains (Fig. 1a).

Analyses of the deduced
A phylogenetic analysis categorized PsANT and its homologs into two distinct groups (Fig. 1b). ANT homologs in fungi, including the single-celled yeast fungus and different multi-cellular pathogenic fungi, were clustered into one large clade, while the homologues in humans and C. elegans were grouped into another clade. Interestingly, the ANT homologues from plants were found to exist in the clade characterized by the fungi members, indicating a closer relationship of ANT in fungi and plants.
PsANT protein localized in mitochondria of the invasion hyphae of Pst. To obtain more information about PsANT in the pathogenic fungi Pst, immuno-cytochemical localization was used to  characterize the functional site of PsANT in Pst. Labeling with the primary antibody prepared from the 61 amino acids (40-100) of PsANT. PsANT was localized to the mitochondria of the infectious hyphae during the compatible interaction of wheat and Pst (Fig. 2), confirming that PsANT is a member of mitochondrial protein. The tight texture of the cells and the regular shape of the mitochondria indicated the active physiological status of the invasion hyphal cells, and the concentration of PsANT in the mitochondria (Fig. 2a,b) may indicate its involvement in energy supply during invasion and hyphal development. Interestingly, PsANT labeling was also found in mitochondria in collapsed cells, in which the cytoplasm was broken down and the mitochondria were aberrant (Fig. 2c,d).
Furthermore, to address the role of PsANT during infection of Pst, we compared the amounts of PsANT in uredinospores, germinated uredinospores, germ tubes and invasion hyphae. The results showed that before invasion, much more abundant labeling was observed in germinated uredinospores than in uredinospores and germ tubes (Fig. 2e), indicating that the energy for germination were mainly supplied by uredinospores and PsANT was responsible for the energy supply due to its translocase activity. After invasion, in the invasion hyphae, more labeling of PsANT was found in mitochondria under normal condition (NM) compared to uredinospres, but less than in germinated uredinospores. This may due to that during germination, the energy supply mainly rely PsANT, and after invasion, Pst also could uptake energy through haustoria from host plants besides depending on PsANT. Interestingly, more PsANT labeling was observed in collapsed mitochondria (CM) than in mitochondria under normal condition. This may indicate that except for energy supply role, PsANT also participate in mitochondria mediated apoptosis.
Over-expression of PsANT leads to cell death in fission yeast with accumulated reactive oxygen species (ROS). Because of the lack of an available transformation system in Pst, we studied the role of PsANT in cell death by the aid of fission yeast system. PsANT was expressed in yeast driven by the nmt promoter, which is repressed by thiamine. Bax and the pREP3x empty vector were used as the positive and negative controls, respectively. Expression of Bax resulted in obvious cell death from 12 h to 16 h, whereas pREP3x did not change fission yeast growth (Fig. 3a). Fission yeast cells expressing PsANT in absence of thiamine (− VB) led to a significant reduced fission yeast cell number compared to those in presence of thiamine (+ VB) from 16 h to 24 h after incubation (Fig. 3a).
To determine whether expression of PsANT or Bax leads to cell death or reduced yeast growth, we checked the cell death phenotype of treated yeast cells with methylene blue staining. As shown in Fig. 3b, through methylene blue staining, the dead yeast cells were stained into blue. Furthermore, we calculated the ration of stained yeast cells in total cells. The results showed that yeast cells expressing PsANT incubated without thiamine (− VB) exhibited significantly higher level of dead cells than that incubated with thiamine (+ VB) from 12 to 20 h (Fig. 3c). Expression of Bax in yeast cells resulted in significant cell death as observed for PsANT (Fig. 3c). The fission yeast cells transformed with empty pREP3x did not show much change in ratio of dead cells. Thus, the results suggested that PsANT induce cell death in fission yeast cells, and the reduced number of fission yeast expressing PsANT or Bax was due to cell death but not reduced growth.
DHR123 staining of H 2 O 2 in yeast cells revealed that from 16 h to 24 h, the oxidant level in PsANT over-expressing yeast cells without thiamine was significantly higher than in those that were incubated with thiamine (Fig. 3d). The fission yeast cells transformed with empty pREP3x did not show much change in ROS accumulation. Thus, the results suggest that the cell death induced by PsANT in fission yeast cells is characterized by accumulation of H 2 O 2 .
Over-expression of PsANT leads to cell death in wheat. Given the close relationgship of ANT in fungi and plants, we addressed whether PsANT could induce cell death in the host of Pst. The single-barreled particle delivery system was used to deliver PsANT and GUS into wheat leaves. The β -glucuronidase (GUS) gene was introduced into tobacco leaf cells as a reporter, and the contribution of the protein encoded by PsANT to cell death was measured by the restoration of cells expressing GUS. When the wheat leaves were bombarded with PsANT + GUS, a 64.2% reduction in the number of GUS blue spots was observed compared to leaves that were shot with empty vector plus GUS (EV + GUS) (Fig. 4a).
The three mito-carr domains contributed differently to cell death induction. As a member of the mitochondria carrier family, PsANT consists of three 100-residue mito-carr repeats. To determine the domains that are responsible for cell death induction, a double-barrel various deletion mutants of PsANT were tested in tobacco leaves, which is easier to operate as a model plant. A particle bombardment assay with a double-barreled attachment was applied to facilitate the direct comparison of test and control bombardments. Expression of PsANT reduced the number of GUS-positive blue patches by 78.7% compared to controls, verifying the cell death induction role of PsANT ( Fig. 4b and Table 1). Then, truncated PsANT mutants were further examined. Transient expression of PsANT 1-110 , PsANT 111-214 or PsANT 215-314 , which contained the first, second or last mito-carr domain, resulted in a 45.9%, 54.7% and 40% reduction of the blue spots, respectively. Meanwhile, transient expression of PsANT 1-214 , which comprised the first two mito-carr domains, led to a decrease of approximately 70% in GUS spots compared  to the control ( Fig. 4b and Table 1). The results indicated that the three mito-carrier domains contribute differently to cell death and likely work together to engage in cell death induction.

PsANT was up-regulated during infection of Pst in wheat. qRT-PCR analyses revealed that
PsANT transcripts were detected at relatively low levels in urediniospores and germ tubes but were sharply induced at 48 hpi and kept constant until 5 dpi (Fig. 5). The results indicate an in-planta induction of PsANT in the infection and propagation stage of the stripe rust fungus.
Silencing of PsANT attenuated Pst growth and development. Due to the lack of a stable genetic transformation system for stripe rust fungi, the host-induced gene silencing (HIGS) technique mediated Values were collected from 10 view fields per treatment and three biological replicates were performed. Data are presented as mean ± SE. Asterisks indicate a significant difference (*P < 0.05, **P < 0.01) from treatment with thiamine using Student's t test.  by BSMV (Barley stripe mosaic virus) was used to knockdown the expression of PsANT in Pst. Two fragments were designed for silencing as indicated in Fig. 6a. Considering the existence of ANT in wheat, we obtained the ANT in wheat variety Shuiyuan 11 (Su11) (Su11-ANT) and compared PsANT with Su11-ANT to examine the specificity of the silencing fragment. PsANT shared a similarity of 65.84% with Su11-ANT at the nucleotide level. No more than 11 consecutive identical nucleotides were observed between the targeted PsANT and Su11-ANT (Fig. S1). This should prevent off-target effects while maintaining adequate complementarity for target gene silencing 25 .
Nine days after inoculation with BSMV, obvious photo bleaching was observed in TaPDS-knockdown plants (Fig. 6b), suggesting that the BSMV-HIGS system worked well. The rust disease phenotypes photographed at 14 dpi showed that Pst with decreased PsANT was still virulent enough to cause a full susceptible phenotype of wheat with no significant change compared to the control plants (Fig. 6c).
qRT-PCR analyses showed that the transcript level of PsANT was knocked down by approximately 69% in the leaves that were infected with BSMV:PsANT compared to that in the BSMV:00-infected leaves (Fig. 6d). The results indicated that PsANT was partially knocked down and confirms that silencing of PsANT did not alter the pathogenicity of Pst.
Silencing of PsANT impeded the growth and development of Pst. Despite of the unchanged disease phenotype, the development and growth of Pst with silenced PsANT were examined through histological observation. The 4 th leaves were sampled at 24 and 48 hpi, when primary haustoria began to form in mesophyll cells and when hyphal growth began. As shown in Fig. 6, the silencing of PsANT decreased the number of haustorial mother cells and haustoria at 24 hpi and 48 hpi ( Fig. 7a-f,j,k). Meanwhile, hyphal growth was also significantly reduced, with shorter hyphae at 24 hpi and 48 hpi and a smaller infection area at 48 hpi ( Fig. 7a-f,l,m). However, by 120 hpi when haustoria formed in great number, the infected area of Pst with silenced PsANT was slightly but not significantly smaller than that of the controls (Fig. 7m).

Discussion
ANT is a physiologically conserved component of programmed cell death in mammals and C. elegans that regulates the permeability of the mitochondrial membrane as a component of the permeability transition pore (PT) 5 . Genetic evidence also suggests that ANT makes a phylogenetic contribution to cell death in another ancient organism, yeast. However, little has been investigated regarding the role of ANT in plant pathogenic fungi. In this study, we identified an adenine nucleotide translocase homologue PsANT in the wheat stripe rust fungus. PsANT contained three typical conserved mito-carr domains of ANT members in yeast and mammals, indicating the conservation of ANT across fungi and animals. Interestingly, different from the multiple ANT members found in other organisms, only one single ANT was discovered in Pst in this study. Since the pathogenic fungus mainly depend on their host for nutrient and energy supply, their own energy supply mechanism might have been degenerated during the evolution, and only one ANT member might have remained to supply energy for pathogen growth before setup of the parasite relationship on host.
To investigate the function of PsANT, we analyzed the location of PsANT in Pst during infection in wheat. Immuno-cytological localization showed that PsANT was concentrated in the mitochondria of the invasion hyphae, confirming that PsANT was a member of the ANT family in the mitochondria. Mitochondrial labeling of PsANT was observed in normal hyphal cells as well as in collapsed cells. ANT is a bi-functional protein that under the normal physiological condition exerts the ATP/ADP translocation role and under apoptotic conditions can be converted into a pro-apoptotic pore under the control of onco-and anti-oncoproteins from the Bax/Bcl-2 family in mammals 26,27 . Thus, combined with the abundant PsANT in germinated uredinospores and the invasion hyphae, the localization of PsANT in mitochondria under the normal condition may be responsible for supplying energy for the germination of uredinospores and development of the invading Pst hyphae through its ATP/ADP translocase activity Whereas, the abundance of PsANT in collapsed mitochondria also indicates that PsANT is involved in apoptosis-like cell death in Pst. Our results provide direct evidence for the mitochondria localization of ANT members in a plant pathogenic fungus, which may coincide with their bi-functional role in energy supply and mitochondria mediated cell death.
Because the lack of a stable transformation system for use in Pst, heterologous system was applied to assess the temporal relationship between PsANT and cell death. Exploiting the yeast system, PsANT was shown to induce cell death as observed for mouse pro-apoptotic Bax. The pro-cell death role of PsANT was consistent with the lethal effect of the three ANT isoforms in yeast 28,29 . Similarly, ANT1 and ANT3 of human were reported to act as pro-apoptosis factors that can dominantly induce apoptosis when over-expressed in human cell lines 10 . Therefore, it is presumable that ANT functions as a conserved cell death component across fungi and mammals. In fact, an earlier comparative analysis of the programmed cell death pathways in filamentous fungi has revealed the existence of conserved core components of the ancestral PCD machinery in Metazoa 21 , which indicated filamentous fungi and Metazoa shared a conserved PCD machinery.

Figure 6. Functional analyses of PsANT during the interaction between wheat and stripe rust using the BSMV-mediated host-induced gene silencing system (HIGS). (a)
The two fragments that were used to silence PsANT. (b) Mild chlorotic mosaic symptoms were observed on the 4 th leaves of the seedlings at 9 dpi with BSMV, and photobleaching was evident on the 4 th leaves of plants that were infected with BSMV:TaPDS. MOCK: Wheat leaves with Fes buffer. (c) Disease phenotypes of the 4 th leaves that were preinoculated with BSMV:00, BSMV:PsANT-as1 or BSMV:PsANT-as2 and then challenged by virulent CYR31. (d) Silencing efficiency assessment of PsANT in the Pst. The 4 th leaves inoculated with BSMV:00 followed by inoculation with CYR31 were used as the controls. The data were normalized to the PsEF-1a expression level. Three biological replicates for each time point were averaged with standard error of mean indicated. Asterisks indicate a significant difference (*P < 0.05) from BSMV:00 using Student's t test.
Scientific RepoRts | 5:11241 | DOi: 10.1038/srep11241 Coping with the close evolution relationship of ANT in fungi and plants, we found that transient expression of PsANT promoted significant cell death in the monocotyledon Triticum aestivum and dicotyledon N. benthamiana. Based on the well established double-barrel bombardment assay in tobacco leaves, we determined the domains of PsANT that are responsible for cell death induction in tobacco. The three mito-carr domains were all required to induce cell death, but made different level of contribution. The second domain made a greater contribution compared to the first and last one. The combination of the first two domains led to more cell death than any single domain but was still not as efficient as the complete PsANT protein. Therefore, we presumed that the three domains worked together to induce cell death. However, a previous study reported that the first three transmembrane helixes of ANT-1 in humans are responsible for inducing apoptosis, and the remaining two transmembrane domains are not required 10 . Considering that a truncated molecule could not form a membrane channel, it was suggested that overexpressed ANT-1 might have triggered cell death through a PT-independent mechanism. The Wheat leaves that were pre-infected with BSMV:00 or recombinant BSMV were followed by inoculation with Pst CYR31 in the compatible interaction. The fungal structure was stained with wheat germ agglutinin (WGA). The fungal growth of Pst pathotype CYR31 in wheat leaves that were inoculated with BSMV:00, BSMV:PsANT-as1 and BSMV:PsANT-as1 at 24 hpi (a-c), 48 hpi (d-f) and 120 hpi (g-i) was observed under a fluorescence microscope. Average number of haustorial mother cells (HMC) (j) and haustoria (k) of Pst in each infection site were counted. (l) Hyphal length, which is the average distance from the junction of the substomatal vesicle and the hypha to the tip of the hypha, was measured by DP-BSW software (unit in μ m). (m) Infection area, the average area of the expanding hypha plus the host cells, was calculated by DP-BSW software (unit in × 10 3 μ m 2 ). All the results were obtained from 50 infection sites and three biological replications were performed. Asterisks indicate a significant difference (*P < 0.05, **P < 0.01) from BSMV:00 inoculated plants using a two tailed Student's t test. SV, substomatal vesicle; HMC, haustorial mother cell; IH, infection hypha; H, haustoria.
Scientific RepoRts | 5:11241 | DOi: 10.1038/srep11241 difference between these findings in humans and our observations of PsANT might suggest that the mechanism by which ANT induces cell death differs between pathogenic fungi and mammals.
The function of ANT has been well studied in mammals and yeasts, but due to the limitations of the yeast system, the role of ANT in pathogen pathogenicity is unclear. The detection of PsANT transcripts in the entire infection stages of Pst suggested that PsANT was of fundamental function to Pst. The in-planta-induced expression of PsANT during the biotrophic stage indicated that PsANT is important for Pst infection, but there is little direct functional evidence to show which genes are involved in rust pathogenicity and virulence because mutagenesis and transformation protocols are lacking in Pst. Recently, host-induced gene silencing (HIGS) was developed as a novel tool to address gene function in obligate biotrophic fungi. The expression in plants of RNA interference (RNAi) constructs targeting fungal genes can specifically silence their targets in the invading pathogenic fungus Blumeria graminis 30 . Gene fragments from the rust fungi Puccinia striiformis f. sp. tritici and P. graminis f. sp. tritici were delivered to plant cells through the Barley stripe mosaic virus (BSMV) system, and some of these fragments reduced the expression of the corresponding genes in the rust fungus 31 . Thus, we studied the likely role of PsANT in the growth and pathogenicity of Pst through knocking down PsANT during the interaction of wheat and Pst using the BSMV-mediated HIGS system. PsANT was efficiently silenced by 40%-60%, and the partial silencing of PsANT led to blocked development and growth of Pst with attenuated formation of haustorial mother cells and haustoria at early stages of infection. Along with this reduction in haustorial formation, hyphal growth and infection were also repressed. In the early stage of infection in host plants, when haustoria have not yet formed in large numbers, Pst depends mainly on the energy supplied by PsANT for invasion and expansion. The results indicate that PsANT plays a role in maintaining the normal development and differentiation of Pst at an early stage through its energy supply function. However, despite the inhibited development at the early stage, the infected area was approximately the same as that of the controls at late stage, suggesting that the pathogenicity of Pst was not fundamentally affected in PsANT-knockdown plants. This may due to that at the late infection stage, haustoria begin to form in great numbers and the expansion of Pst mainly depends on the nutrient and energy uptake from host cells through haustoria. Thus, the silencing of PsANT did not substantially influence Pst infection during the late stage. These results indicate that PsANT contributes to fungal growth and development probably due to its energy supply role during early infection but is not essential for virulence. This possibly could further explain the lower copy numbers of ANT in Pst than in its host wheat and other speicies.
In conclusion, our study demonstrated that PsANT promotes cell death and contributes positively to the growth and development of Pst. These findings suggest that fungi, plants and animals shared phylogenetically conserved components in the cell death mechanism. Further studies addressing how PsANT modulate fungal cell death will be required to share light on the concrete cell death mechanism of pathogenic fungi. Notwithstanding that, the study of fungal cell death and pathogenicity mechanism will be of great help in designing of fungicides for controlling the disease.

Methods
Plant materials, strains and growth conditions. Puccinia striiformis f. sp. tritici (Pst) pathotype CYR31 and wheat genotype Su11 were used in this study. Pst race CYR31 was propagated on Su11 as described by Kang &Li 32 . Fresh urediniospores were harvested from infected wheat leaves and dispersed uniformly in sterile distilled water. After incubation at 9 °C for 10 h, germinated urediniospores were collected, quickly frozen in liquid nitrogen and stored at − 80 °C for RNA extraction. To study the expression of PsANT during infection, wheat leaves that were infected with Pst CYR31 were sampled at 24, 48, 72, 120 and 168 h post-inoculation (hpi). For each assay, three biological replicates were applied.
Nicotiana benthamiana plants, which were used for transient over-expression by particle bombardment, were grown at an ambient temperature of 25 °C under a light cycle of 16 h light/8 h dark.
Total RNA extraction and qRT-PCR. Total RNA of urediniospores, germ tubes and wheat leaves that were infected with Pst were isolated using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA). A 2.0-μ g RNA aliquot of each sample was used for cDNA synthesis with an Oligo(dT) 18 primer using the Reverse Transcription PCR system (Promega, Madison, WI, USA). The resulting 1st-strand cDNA products were used as the templates for qRT-PCR. qRT-PCR was conducted as described by Wang et al. 33 with a 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The Pst elongation factor Pst_EF gene 31 was used as the internal reference for each qRT-PCR reaction. All of the reactions were performed in triplicate, and reactions without templates were used as negative controls. The comparative 2 -ΔΔCT method was used to quantify relative gene expression 34  Isolation of cDNA and sequence analysis. The primers PsANT _F and PsANT _R were designed on the basis of EST sequence WRIS_455 (GenBank # GR304380) from a cDNA library that was constructed from wheat that was infected with the virulent Pst pathotype CYR31 35 . The open reading frame (ORF) was amplified from the 1 st -strand cDNA that was synthesized with RNA from the urediniospores of Pst race CYR31.
The deduced amino acid sequence of PsANT was analyzed with PFAM for conserved domains and motifs. TMPRED was used to predict the trans-membrane domains. Multiple sequence alignments were Scientific RepoRts | 5:11241 | DOi: 10.1038/srep11241 carried out using DNAMAN software, and phylogenetic analysis of PsANT and other ANT members was performed using the neighbor-joining method.

Plasmids construction. The plasmid pUCPsANT that was used in transient expression in tobacco
and wheat leaves was constructed as described by Dou et al. 36 PUCPsANT (PsANT driven by the CaMV 35S promoter) was obtained by inserting the coding sequence of PsANT into Xma I-and Kpn I-digested plasmid pUCBAX (BAX gene driven by the CaMV 35S promoter). Truncated mutants of PsANT, including pUCPsANT  , pUCPsANT 111-214 and pUCPsANT 215-314 , which contained the first, second and last mito-carrier domains, respectively, were generated in a similar manner, as was the pUCPsANT 1-214 construct, which represented the first two conserved mito-carrier domains.
To generate the PsANT construct that was used for bacterial expression in E. coli, primers were designed to amplify a 186-bp fragment with a stop codon, which encodes the amino acid sequence of PsANT from residues 40 to 100. The 186-bp fragment was inserted into pET-32a(+ ) vector (Novagen) that had been digested with BamHI and HindIII.
For the overexpression assay in yeast, the coding sequence of PsANT was amplified. The PCR products were digested with XhoI and BamHI and inserted into the digested pREP3x vector (driven by repressed nmt promoter) 37 . pREP3x_Bax was constructed by inserting the coding sequence of Bax into XhoI-and SalI-digested pREP3x vector.
A BSMV γ RNA-based vector was constructed as described by Holzberg et al. 38 The cDNA fragments derived from the coding sequence and the 3' untranslated region, which were 190 bp (949-1138) in size, and from the coding sequence, which were 202 bp (131-332) in size, were used to construct the recombinant PsANT-as1 and PsANT-as2 plasmids, respectively. To guarantee the specificity of gene silencing, the fragments that showed the highest polymorphism within the gene family and the lowest sequence similarities with other genes using a BLASTN search in the NCBI database were chosen to construct γ RNA-based derivative plasmids.
The primers for all of the plasmid constructions are documented in Supplementary Table S1.
Immuno-cytochemical localization. The prokaryotic expression of the PsANT protein in E.
coli strain BL21 (DE3) with the generated construct and the preparation of the primary polyclonal antibody in rabbit were completed by Hangzhou Anhua Biotechnology Co., Ltd., China. For the immuno-cytochemical localization of PsANT, the samples were prepared and analyzed as previously described 39 . For immuno-electron microscopy, the primary leaves of wheat that was infected with Pst race CYR31 were harvested 5 dpi. Small leaf specimens (~2 mm 2 ) were excised from the infected tissues and fixed with 3% glutaraldehyde (v/v) in 100 mM phosphate buffer (pH 6.8) for 3 h at 4 °C. After washing in phosphate buffer four times (15 min each), the leaf sections were post-fixed with 1% (w/v) OSO4 in 100 mM phosphate buffer (pH 6.8) and again washed four times for 15 min each. The samples were then dehydrated in a graded ethanol series and embedded in LR-white resin (Berkshire, U.K.). The embedded samples were cut with a vibratome (TPI, Series 1000) to obtain ultra-thin sections. After incubation at 9 °C for 6 h, germinated urediniospores were collected. The samples for immuno-cytochemical localization of Pst uredinspore and germinated uredinospores were prepared as described for the wheat leaf samples. The sections were blocked three times for 20 min each with 1% (w/v) BSA in PBS (50 mM Na-phosphate, 150 mM NaCl, pH 7.2) and incubated with the rabbit anti-PsANT antibody (1:1000) in 1% BSA for 2 h at room temperature. After three washes in 1% BSA (15 min each) and two washes in PBS, the sections were incubated with the secondary colloidal gold antibody (goat anti-rabbit IgG conjugated to 15 nm gold particles (1:20) diluted with PBS for 1 h at room temperature. The samples were post-stained with 5% (w/v) uranyl acetate in 50% (v/v) ethanol for 8 min and in lead citrate for 6 min. The sections were examined with a Zeiss-EM10 transmission electron microscope (Tokyo, Japan) at an 80-kV accelerating voltage.
Transient expression of PsANT in tobacco and wheat by particle bombardment. The leaves of 5-7-week-old N. benthamiana plants were used for the transient expression of PsANT. Over-expression of PsANT was achieved through particle bombardment assays using the Bio-Rad He/1000 particle delivery system with an attached double-barreled extension 36 . For bombardment, 100 μ L of 90 mg/mL tungsten particles was combined with 50 μ g of empty vector and 50 μ g of GUS plasmid (pUCGUS) or a mixture of 50 μ g of pUCPsANT and 50 μ g of pUCGUS. The prepared tungsten-DNA mixture was used for 30 shots. To quantify the role of PsANT in cell death, empty vector (1.67 μ g/shot) and pUCGUS (1.67 μ g/ shot) were bombarded onto half of the tobacco leaf in one barrel as controls. In the second barrel, the pUCPsANT plasmid (1.67 μ g/shot) and pUCGUS (1.67 μ g/shot) were delivered into the other half of the tobacco leaf. After incubation at 28 °C for 3 d in darkness, the bombarded leaves were stained for 16 h at 28 °C using 0.8 mg/mL X-gluc (5-bromo-4-chloro-3-indolyl-b-D-glucuronic acid, cyclohexylammonium salt); 80 mM Na phosphate, pH 7.0; 0.4 mM K 3 Fe(CN) 6 ; 0.4 mM K 4 Fe(CN) 6 ; 8 mM Na 2 EDTA; 0.8 mg/ mL 20% methanol; and 0.06% (v/v) Triton X-100 and then de-stained in 100% ethanol. For each shot pair (GUS+ PsANT DNA versus GUS+ control DNA), the logarithm of the ratio of the blue spots on the leaves that were bombarded with GUS and PsANT to that of the control barrels was calculated using the Wilcoxon signed ranks test. At least 16 shot pairs were performed.
For the over-expression of PsANT in wheat leaves, the first leaves of 7-day-old wheat Su11 were used in the particle bombardment assay. The wheat leaves were packed tightly into dishes and shot with the Bio-Rad He/1000 single-barreled particle delivery system according to the previously described protocol 40,41 . All of the plasmids that were used were prepared at a concentration of 1 μ g/μ L. For bombardment, 26.7 μ L of 90 mg/mL tungsten particles was mixed with 7 μ g of the GUS plasmid (pUCGUS) and 7 μ g of the empty vector or 7 μ g of pUCPsANT, after which an equal volume of 1 M Ca(NO3) 2 was added. After incubation for 10-30 min at room temperature with periodical inversion, the mixtures were centrifuged at 14,000 rpm for 30 s. Then, the pellets were successively washed with 1 mL of 70% ethanol and 1 mL of 96% ethanol. Finally, the pellet was re-suspended in 30 μ L of 96% ethanol. For each shot, 4 μ L of the prepared DNA-tungsten mixture was applied. The leaves were kept at 28 °C in darkness for 2 days and then stained as described above for 16 h and de-stained in 100% ethanol. The blue shots that formed on the leaves that were shot with GUS and PsANT plasmids were compared to those that formed on the leaves that were shot with GUS and empty vector. Each assay consisted of seven shots and was conducted at least twice. Significant differences between the treatment and controls were analyzed with a two tailed Student t-test.
Several truncated mutants of PsANT were constructed and tested in tobacco to identify the functional domain of PsANT. pUCPsANT 1-110 and pUCPsANT 1-214 excised the last two mito-carrier domains and the last mito-carrier domain, respectively. pUCPsANT 111-214 and pUCPsANT 215-14 contained the middle mito-carrier domain and the last mito-carrier domain, respectively.
Over-expression of PsANT in fission yeast. The reconstructed vector pREP3x_PsANT, pREP3x _Bax and pREP3x empty vector were transformed into Schizosaccharomyces pombe by electroporation. Thiamine was used as the repressor of the nmt promoter in the pREP3x vector at a concentration of 5 μ g/ mL. Positively transformed cells were verified by PCR with pREP3x_F and pREP3x_R primers. The positively transformed cells were incubated in EMM medium with a starting optical density of OD 600 = 0.01 for 24 h with or without thiamine. The incubated fission yeast cells were sampled at 12,14,16,18,20,22 and 24 h post incubation, and the fission yeast cells were counted using a hemocytometer. The cell death phenotype of treated yeast cells were checked by methylene blue (Sigma-Aldrich, St. Louis, MO) staining as described by Fraser 42 . The dead yeast cells were stained into blue and the ratio of dead yeast cells in total cells were calculated from at least 10 view fields. The levels of mitochondria oxidants in fission yeast cells were determined by dihydrorhodamine 123 (DHR123) staining as previously described 43 . The stained yeast cells were observed under an Olympus BX-51 fluorescence microscope (Olympus Corp., Tokyo) and measured from at least 10 view fields. The proportion of yeast cells with ROS accumulation was calculated. For all the experiments three biological replications were performed.
BSMV-mediated silencing of PsANT in the compatible wheat-Pst interaction. Capped transcripts were prepared from the linearized plasmids containing the tripartite BSMV (Barley stripe mosaic virus) genome using the mMESSAGE T7 in vitro transcription kit (Ambion Austin, TX, U.S.A.) following the manufacturer's instructions. The second leaf of the wheat seedlings at the two-leaf stage was inoculated with BSMV transcripts by gently rubbing the surface with a gloved finger 38,44 . The BSMV-infected plants were maintained in a growth chamber at 23 ± 2 °C. Nine days after BSMV infection, the fourth leaves were inoculated with fresh urediniospores of the Pst pathotype CYR31. The disease symptoms of the fourth leaves were recorded at 14 days post-inoculation (dpi) with rust fungi. Three independent sets of wheat plants were prepared for each of the four BSMV constructs (BSMV:00, BSMV:PsANT-as1, BSMV:PsANT-as2 and BSMV:TaPDS) with a total of 72 seedlings. TaPDS encodes the Triticum aestivum phytoene desaturase, which has been used as control of virus-induced gene silencing 44 . Another 18 seedlings were mock inoculated with 1× Fes buffer 45 and used as the control. Three independent biological replicates were performed.
The fourth leaves were sampled at 24, 48 and 120 hpi for qRT-PCR and histological observation. The silencing efficiency of PsANT was confirmed using qRT-PCR as described above.

Histological observation of fungus growth.
To determine the effects of silencing PsANT in stripe rust fungus, fungal development was examined with a microscope. The sampled wheat leaves were fixed and stained as previously described 46 . Cleared segments were examined with an Olympus BX-51 microscope (Olympus Corp., Tokyo) for infection sites. Only an infection site on which substomatal vesicles had formed in the stomata was considered a successful penetration. At least 50 infection sites were examined for each of the five randomly selected leaf segments for each treatment. Wheat germ agglutinin (WGA) conjugated to Alexa 488 (Invitrogen, Carlsbad, CA, USA) was used to stain the infection structure of Pst as previously described 47 . The hyphal length, hyphae branches and haustorial mother cells were observed and calculated using DP-BSW software. Standard deviations and Student's t-test were applied for statistical analysis.