A Xanthomonas oryzae type III effector XopL causes cell death through mediating ferredoxin degradation in Nicotiana benthamiana

Plant pathogenic bacteria engage in an ongoing struggle for survival. Gram negative phytopathogenic prokaryotes deploy type III secretion system to deliver effector proteins into plant cells to manipulate the host cellular environment. In many cases, the targets of these effector proteins are unknown. The bacterial blight pathogen, Xanthomonas oryzae pv. oryzae (Xoo), causes a devastating disease in rice and the Xoo-rice interaction is an example of the ongoing evolutionary battle between plants and pathogens. In this study, we show that an effector XopL from Xoo exhibits E3 ubiquitin ligase activity and induces cell death in the nonhost Nicotiana benthamiana (Nb). Through yeast two-hybrid, co-immunoprecipitation, and biomolecular fluorescence complementation assays, it was shown that XopL interacts with ferredoxin (NbFd), an electron transport molecule in plant chloroplasts. XopL induced defense-related response in Nb and specifically targeted NbFd for ubiquitination and ubiquitin-mediated degradation, a process that increased the production of reactive oxygen species (ROS). This study demonstrates the ubiquitination and degradation of NbFd as a result of XopL activity, which further illustrates the tremendous functional diversity in Xanthomonas effector proteins.


Background
Plants are continually bombarded by abiotic and biotic stressors, including phytopathogenic viruses, fungi, nematodes, and bacteria. The bacterial genus Xanthomonas includes over 30 pathogenic species that cause diseases on many economically-important crops including banana, Brassica spp., cassava, citrus, rice, tomatoes, and tobacco (Schornack et al. 2013;Jacques et al. 2016). Protein secretion systems (e.g. Types I-VI, Sec, TAT) play important roles in the successful colonization of host plants by phytopathogenic bacteria, including Xatnhomonas spp. (Kay and Bonas 2009;Büttner 2016). Many pathogenic xanthomonads rely on the type III secretion system (T3SS) to successfully invade and multiply inside plant tissues (Kay and Bonas 2009). The T3SS translocates type III effector (T3E) proteins into host cells, which can facilitate pathogen-mediated manipulation of plant cellular pathways and subsequent disease or trigger plant defense responses to thwart the infection process (Dean 2011;Lee et al. 2013).
During co-evolution with pathogens, plants developed a multi-layered immune system to provide protection from pathogen ingress (Dodds and Rathjen 2010). The first layer of defense involves the recognition of pathogen-associated molecular patterns (PAMPs), such as bacterial flagellin, peptidoglycan, and chitin. The recognition of PAMPs is modulated by plant-encoded pattern-recognition receptors (PRRs), which include receptor-like kinases and proteins that are often located in plant cell membranes. The recognition of PAMPs results in defense known as PAMP-triggered immunity (PTI) and the activation of responses that interfere with pathogen invasion and colonization (Schwessinger and Ronald 2012;Wu and Zhou 2013). In addition to PTI, plants have evolved another layer of defense known as effector-triggered immunity (ETI); this is activated via the direct or indirect recognition of pathogen effectors by plant resistance (R) genes and is associated with the hypersensitive response (HR), a form of programmed cell death that limits pathogen multiplication (Wu et al. 2014). Thus, T3Es may either trigger the plant defense response (Khan et al. 2016) or manipulate plant cellular pathways in ways that benefit the bacterial pathogen (Büttner 2016).
Plants also evade many potential pathogens due to nonhost resistance (NHR), which is the most common form of plant resistance and is effective against a wide array of pathogens (Mysore and Ryu 2004;Fan and Doerner 2012). The mechanism of NHR is complex, which includes preformed physical barriers, secondary metabolites, and induced defense. Plant NHR reactions vary considerably, ranging from symptomless reactions to the HR and programmed cell death (Mysore and Ryu 2004;Uma et al. 2011;Senthil-Kumar and Mysore 2013). In some cases, bacterial effectors are recognized by nonhost plants, thus triggering an HR or cell death. For example, XopQ (HopQ) from Xanthomonas triggers ETI on the nonhost Nb (Wei et al. 2007;Schwartz et al. 2015). Recently, a resistance protein from Nb designated Roq1 (Recognition of XopQ1) was shown to interact with XopQ, which led to an HR (Schultink et al. 2017). Moreover, the transgenic expression of Roq1 in beet plants was sufficient to trigger recognition of XopQ, thus indicating that R genes from heterogeneous plant species may improve host resistance in unrelated species (Schultink et al. 2017).
The closely-related xanthomonads, X. oryzae pv. oryzae (Xoo) and X. oryzae pv. oryzicola (Xoc), cause bacterial blight (BB) and bacterial leaf streak (BLS) on rice, respectively (Nino-Liu et al. 2006;Zhang and Wang 2013). Like many other bacterial pathogens, successful infection by X. oryzae depends on the T3SS, which is necessary for its pathogenicity on host rice and the HR on nonhost Nb. Two kinds of proteins traverse the T3SS in X. oryzae; these include the T3E proteins that are directly delivered into plant cells (Kay and Bonas 2009) and extracellular accessory proteins such as the harpin Hpa1 (Zou et al. 2006).
Xanthomonas T3Es are broadly categorized into two different families: TALE (transcription activator-like effectors) and non-TALE which are also called Xanthomonas outer proteins (Xop) effectors. Many Xop effectors interfere with PTI and ETI-associated plant immunity (Büttner 2016), however, only relatively few of them are well-defined in the precise cellular targets and modes of action. For example, XopP targets OsPUB44 U-box and inhibits rice immunity (Ishikawa et al. 2014). XopY (Xoo1488) inhibits rice immunity by suppressing OsCERK1-mediated phosphorylation of OsRLCK185 (Yamaguchi et al. 2013).
Recently, ubiquitination was shown to be utilized by pathogen effectors to gain a competitive edge (Ustun and Bornke 2014;Pruneda et al. 2016;Lin and Machner 2017). Ubiquitination alters the fate of a target protein, often tagging it for subsequent degradation by the 26S proteasome (Ustun and Bornke 2014). A number of Xanthomonas T3Es interact or modulate the host ubiquitin-proteasome system including XopJ, XopD, XopK, XopAE, and XopL (Singer et al. 2013;Ustun and Bornke 2014;Pruneda et al. 2016;Popov et al. 2018;Qin et al. 2018). Singer et al. demonstrated that XopL from X. campestris pv. vesicatoria (XopL Xcv ) contains E3 ubiquitin ligase activity and interacts with plant E2 ubiquitin conjugating enzymes (Singer et al. 2013). Recently, the transient expression of XopL Xcv in Nb caused major ultrastructural changes including the relocation of plastids to the nucleus (Erickson et al. 2018). Despite these advances in the understanding of XopL function, the precise targets of XopL have remained unclear.
In this study, we explore the role of XopL Xoo in mediating nonhost resistance in Nb, a model plant that has been widely-used to study the function of numerous Xanthomonas T3Es including XopH, XopJ, XopQ, XopX and XopL (Thieme et al. 2007;Singer et al. 2013;Stork et al. 2015;Adlung et al. 2016;Blüher et al. 2017;Erickson et al. 2018). A yeast two-hybrid assay was used to demonstrate the interaction between XopL and ferredoxin (NbFd). The interaction of XopL and NbFd was further confirmed by co-immunoprecipitation and biomolecular fluorescence complementation assays. XopL was shown to catalyze the ubiquitination of NbFd, which led to subsequent degradation of NbFd in planta and increased levels of reactive oxygen species and defense genes.

XopL induces cell death in Nicotiana benthamiana
X. oryzae triggers strong cell death in non-host N. benthamiana (Nb). In order to find new cell death inducer(s) from X. oryzae, an initial screen for effectors that induce cell death in Nb was conducted in X. oryzae pv. oryzicola (Xoc) RS105. Twenty-two predicted T3Es (http://www.xanthomonas.org/t3e.html) from Xoc were amplified using primers in the Additional file 1: Table  S1 and transiently expressed via Agrobacterium in Nb leaves. In addition to the positive harpin Hpa1 (Zou et al. 2006), five effectors, namely, XopL Xoc , XopP1 Xoc , XopP2 Xoc , XopAA Xoc and XopY Xoc , triggered cell death when transiently expressed in Nb (Fig. 1).
According to the Xanthomonas resource website (www.xanthomonas.org/t3e.html) and previous work (Singer et al. 2013), XopL is conserved in nine out of thirteen Xanthomonas strains from eight species. Thus, we cloned xopL from Xoo model strain PXO99 A and found that XopL Xoo , like XopL Xoc and XopL Xcv (Singer et al. 2013), also induced cell death in Nb, which indicates this function is conserved in Xanthomonas species. Given the fact that the bacterial blight pathogen Xoo PXO99 A is a model strain well-studied, we performed a study on XopL Xoo (XopL hereafter). XopL induced cell death at 7 days post-inoculation (dpi) when transiently expressed in Nb (Fig. 2a). Evans blue staining and ion leakage measurements confirmed that XopL triggered cell death in Nb ( Fig. 2b and Additional file 2: Figure  S1). Localization of XopL:YFP fusions indicated that XopL was primarily localized to the cytoplasm (Additional file 3: Figure S2).
To determine whether XopL inducing-cell death was accompanied with defense responses in Nb, we examined the expression of several defense-related genes in Nb agroinfiltrated with pHBxopL and pHB (empty vector control). Highly expressed genes include the pathogenesis-related (PR) genes PR4, PR5 and PR10, the HR marker gene HIN1 (Gopalan et al. 1996), the master immune regulatory gene NPR1, PAMP-triggering immunity-related gene PTI5 (Nguyen et al. 2010) and elicitor-inducible cytochrome P450 CYP71D20 (CYP) (Lacombe et al. 2010) (Fig. 2c). Furthermore, the expression of xopL resulted in ROS (H 2 O 2 ) accumulation; this was obvious at 3 dpi and was more intense at 6 dpi (Fig. 2d,e). Thus, XopL induces cell death when transiently expressed in Nb, and this is accompanied by the defense genes activation and ROS accumulation.

XopL interacts with Nb ferredoxin protein
To better understand the mechanistic basis for XopLmediated defense in Nb, XopL-interacting proteins were identified from an Nb cDNA library (Sun et al. 2015) of mature leaves using full-length XopL as bait (pGBKT7-xopL, Additional file 4: Table S2) in a yeast two-hybrid assay. A total of 56 positive colonies were initially identified, which represented 11 different genes. To further confirm positive interactions, we swapped the bait and prey roles. The xopL gene was cloned into pGADT7 and used as prey; the coding regions of the 11 different genes were cloned into pGBKT7 as baits and co-transformed into S. cerevisiae AH109 with pGADT7-xopL. Clones encoding ferredoxin (NbFd, AKQ98361) but not empty BD interacted with XopL ( Fig. 3b and Additional file 5: Figure S3). According to co-immunoprecipitation (CoIP) results, we found XopL co-immunoprecipitated with NbFd-YFP but not YFP (Fig. 3c), which suggests an interaction between NbFd and XopL in planta.
We conducted BiFC assays to further explore the interaction between XopL and NbFd. Confocal microscopy indicated that the co-expression of XopL:YN and NbFd:YC generated yellow fluorescent signals in the cytoplasm (Fig. 3d, top row, YFP panel), which were absent in the controls. These results confirmed that XopL interacts with NbFd in the cytoplasm.
Ferredoxin is encoded by nuclear DNA and transported into chloroplast via the chloroplast addressing peptide at the N-terminus end (Pilon et al. 1992). Localization assays of the NbFd:YFP fusion indicated T3Es were transiently expressed in tobacco leaves mediated by Agrobacterium. Harpin protein Hpa1 was used as positive cell death inducer. +++, ++, and + indicate cell death appeared at 4, 7, and 10 dpi, respectively. b Cell death phenotypes induced by effectors. Picture was taken at 10 dpi that most of the fluorescence appeared in chloroplast (Additional file 3: Figure S2, blue arrows); however, faint fluorescent signals were also observed in the cytoplasm (Additional file 3: Figure S2, red arrows), presumably during the transport of NbFd to the chloroplast. NbFd contains the N-terminal signal peptide for import into the chloroplast (residues 1-51) (Holbrook et al. 2016), whereas the C-terminal functional domain contains the conserved [2Fe-2S] iron-sulfur active center and the four conserved cysteine residues that catalyze electron transfer (Hanke and Mulo 2013) (Fig. 3a). To clarify which region of NbFd interacts with XopL, we deleted the chloroplast-addressing peptide of NbFd, generating NbFdΔ2-51. NbFdΔ2-51 interacted with XopL ( Fig.  3b), indicating that the XopL-interacting region of NbFd is likely located in the C-terminal portion. These results were supported by BiFC analysis, in which XopL:YN and NbFdΔ2-51:YC were co-expressed; the interaction of XopL:YN and NbFdΔ2-51-YC resulted in a strong fluorescent signal (Fig. 3d, second row, YFP panel).
XopL Xoo shares 74.24% amino acid identity with XopL Xcv (Additional file 6: Figure S4) that contains Nterminal leucine-rich repeats (LRR) and a C-terminal E3 ubiquitin ligase domain designated as the XL-box (Singer et al. 2013); these regions are also highly conserved in XopL Xoo ( Fig. 3a; Additional file 6: Figure S4bd). To determine which region(s) of XopL interacts with NbFd, we designed constructs containing either the LRR or XL domain of XopL. Both the two domains of XopL interacted with NbFd, however, yeast strains coexpressing XopL-LRR/NbFd or XopL-XL/NbFd failed to grow on SD/−Leu-Trp-His-Ade medium when diluted 1000-fold (Fig. 3b). Our results suggest that NbFd interacts with both the LRR and XL domains of XopL; however, the affinity of NbFd to these two domains is lower than that to the full-length of XopL (Fig. 3b).

XopL mediates the ubiquitination and degradation of NbFd
The affinities of XopL, and the E3 ligase domain in XopL for NbFd led us to investigate whether XopL ubiquitinates NbFd. To test this hypothesis, an in vitro ubiquitination assay was conducted using the Ubiquitin Conjugation Kit K-960 (Boston Biochem, Cambridge, MA, USA). Anti-GST antibody was used to indicate the presence of GST:XopL protein. Results showed that single band was detected in each reaction system containing GST:XopL (Fig. 4a, bottom channel; Additional file 7: Figure S5), which means that no modification occurred in GST:XopL. Then anti-ubiquitin antisera was used to detect the polyubiquitination reaction. Polyubiquitin smear ≥90 kDa was present in lanes 1 and 2, while was absent from lanes lacking the E1 + E2 enzymes (Fig. 4a, middle panel), indicating that, in combination with E1 and E2, XopL could catalyze the formation of polyubiquitin molecules. When probed with anti-His antibody to detect NbFd:His protein, multiple large bands (≥90 kDa, see immuno-reactive smear) were detected only in Next, we investigated whether XopL mediates the degradation of NbFd in planta by transiently co-expressing NbFd:YFP and 3xflag:XopL or empty vector (pHB). By comparing the blotted NbFd:YFP intensity, we found that co-expression with XopL led to reduced accumulation of NbFd:YFP at 2, 3 and 4 dpi, with the lowest accumulation amount at 4 dpi (Fig. 4b, top panel). The NbFd levels were similar when NbFd:YFP was incubated with pHB ( Fig. 4b, top panel, lanes 1, 3 and 5). The levels of actin, the internal control, remained similar regardless of the presence or absence of XopL (Fig. 4b, bottom panel). These results suggested that XopL modulates degradation of NbFd. However, the levels of NbFd remained unaltered in the presence of XopL-LRR, regardless of time point (Fig. 4c, top panel). To exclude the possibility that differences in mRNA levels resulted in altered protein levels, we checked the transcription of NbFd-YFP in infiltrated regions of Nb leaves by RT-PCR tests. Both NbFd-YFP and EF1α (endogenous control) expression levels were similar at each time point (Fig. 4b, NbFd-YFP and EF1α panels), indicating that the difference in NbFd:YFP accumulation was not due to differential expression. To confirm that NbFd was degraded by the ubiquitin-proteasome system, we utilized MG132, a proteasome inhibitor that can reduce the degradation of ubiquitin-conjugated proteins. When MG132 was infiltrated into Nb leaves expressing NbFd-YFP and XopL, NbFd degradation was inhibited (Fig. 4d, ab-myc panel). Collectively, our results imply that XopL mediates the ubiquitination and subsequent degradation of NbFd.
In addition, we observed large immuno-reactive XopL and XopL-LRR bands when probed with anti-flag antibody at 2, 3 and 4 dpi in Nb (Fig. 4b, c, second panel,  lanes 2, 4 and 6), which suggested that a modified form(s) of XopL was also present.

Silencing of NbFd results in ROS accumulation
To further investigate XopL-mediated cell death in Nb and the potential role of NbFd, we designed construct pCA-2 × NbFd for RNAi-mediated silencing of NbFd (Fig. 5a). DNA sequence blasting against the Nb database housed in SolGenomics (https://solgenomics.net) revealed 4 NbFd-encoding genes sharing high identity (98-99%). The region of NbFd selected for silencing contained 33 bp of the chloroplast peptide and 368 bp of the functional domain (Fig. 3a) and was able to match all of the four NbFd coding genes. The NbFd-silenced plant lines showed a dramatic decrease in NbFd expression (approximately 3% of EV1300 lines, transformed with empty vector) in the T 1 generation (Fig. 5b). Furthermore, the silencing of NbFd resulted in mottled, chlorotic leaves (Fig.  5c). Ferredoxin is an ion carrier in chloroplasts and plays an important role in maintaining redox homeostasis in the organelle (Hanke et al. 2004); thus a sharp decrease in ferredoxin levels could impact the redox state of cells and potentially cause ROS accumulation. To investigate this further, we examined H 2 O 2 accumulation in NbFd-silenced Nb leaves. Results showed that regions of NbFd-silenced leaves were stained brown (Fig. 5e) indicating an accumulation of ROS. Whereas, this staining did not occur in EV1300 leaves.
Next, we evaluated the induction of cell death in NbFd-silenced plants. At 3 dpi, both XopL and pHB induced cell death in NbFd-silenced leaves; however, no cell death was observed in leaves of EV1300 infiltrated with pHB in the inoculated region at 7 dpi (Fig. 5d). These results suggest that the silencing of NbFd in Nb renders cell death, which was apparent when both pHB and pHB-xopL were introduced.

ROS accumulation is a prerequisite for XopL-induced cell death
Given that XopL mediated the degradation of NbFd (Fig.  4b) and induced ROS accumulation (Fig. 2d), we speculated that ROS accumulation may be important for XopLmediated cell death in Nb. In order to investigate this, we co-infiltrated pHB-XopL with NAC (N-acetyl-L-cysteine), which was used as an ROS scavenger (Valassakis et al. 2018). NAC inhibited both ROS accumulation (Fig. 6a) and XopL-mediated cell death (Fig. 6b). These results indicate the vital role of ROS accumulation in XopLinduced cell death. Fig. 4 XopL mediates the ubiquitination and degradation of NbFd. a In vitro assay for XopL-mediated ubiquitination of NbFd. Anti-His antisera (ab-His) was used to detect His-tagged NbFd or its ubiquitinated products, anti-ubiquitin antibody (an-ub) was used to detect ubiquitin or polyubiquitin, and anti-GST antibody (ab-GST) was used to detect XopL. The experiment was repeated three times with similar results. b XopL mediates NbFd degradation in planta. NbFd was detected with anti-myc antibody at 2, 3 and 4 dpi. XopL was detected using anti-flag antisera, and total protein was monitored with anti-actin antibody. mRNA of transiently expressed NbFd-YFP and native EF1α were measured at each time interval by RT-PCR. c LRR domain of XopL fails to mediate NbFd degradation. NbFd:YFP and pHB-xopL-LRR (lacking E3 ligase domain) or empty pHB were agroinfiltrated into Nb. NbFd, XopL-LRR, and actin levels were measured by immunoblotting as described in panel b. d Effect of the proteasome inhibitor MG132 on the stability of ferredoxin. Anti-flag and actin antibodies were used to detect flag-tagged XopL and actin respectively. In panels b, c, and d, the values above each blotted band represent band intensity and were calculated using Image J software. In each panel, the first lane band intensity was normalized as 1.00. Experiments were repeated at least three times with similar results

Discussion
Ferredoxins comprise a group of small [2Fe-2S]-containing proteins that are encoded in the nucleus. Fds are synthesized in the cytosol as precursors and are translocated across the chloroplast envelope via an N-terminal extension called the signal (transit) peptide (Pilon et al. 1992;Holbrook et al. 2016). In the current study, NbFd: YFP was observed in both chloroplasts and cytosol (Additional file 3: Figure S2), which is consistent with the intracellular transport process of the Fd pre-protein (Pilon et al. 1992). The interaction between NbFd and XopL is in cytoplasm (Fig. 3d), which suggests that Values represent mean ± SD (n = 3). EF1α was used as a reference gene. c Phenotypes of EV1300 and NbFd-silenced plants. The three uppermost leaves were photographed from representative EV1300 and NbFd-silenced plants. d Cell death on leaves of EV1300 and NbFd-silenced plants in response to XopL and pHB. Agrobacterium strains harboring pHB-xopL or empty pHB were inoculated; pictures were taken at 7 dpi. e ROS (H 2 O 2 ) accumulation in 6-week-old EV1300 and NbFdsilenced leaves using DAB staining. ROS accumulation in cells was visualized by light microscopy with a 20× objective Fig. 6 Effect of the ROS scavenger N-acetyl-L-cysteine (NAC) on XopL-induced cell death. Agrobacterium strains harboring pHB-xopL or empty pHB were inoculated into the wild-type Nb leaves at OD 600 = 1.0; NAC was co-infiltrated at a concentration of 2 mM. a Effect of NAC on XopLinduced ROS accumulation at 3 and 6 dpi. Agrobacterium-infected Nb leaves were collected at 3 and 6 dpi and stained in DAB solution. b Phenotypes of Nb leaves infiltrated with pHB-XopL and pHB, with or without NAC. Pictures were taken at 7 dpi. Experiments were repeated three times with similar results XopL-induced defense related response may be due to NbFd imbalance in chloroplast. Depletion of NbFd as an electron acceptor would enhance production of ROS and ion leakage of cells (Lintala et al. 2012). Chloroplast-generated ROS have been closely associated with cell death (Liu et al. 2007;Zurbriggen et al. 2009) and serve as messengers to initiate the expression of defense genes (Shapiguzov et al. 2012). The C-terminus of NbFd (NbFdΔ2-51) contains the iron-sulfur site and cysteine residues that catalyze electron transport from photosystem I to acceptors located in the chloroplast stroma (Hanke and Mulo 2013). Although lacking the signal peptide for translocation into the chloroplast, NbFdΔ2-51 interacted with XopL in yeast two-hybrid and BiFC assays (Fig. 3b, d). We did not evaluate whether XopL interacts with the signal peptide of NbFd, so it remains unclear whether they interact with each other.
Ubiquitination is a posttranslational modification involving the covalent addition of a single ubiquitin molecule or a polyubiquitin chain onto a lysine residue of the target protein (Kerscher et al. 2006). In this study, we show that XopL Xoo ubiquitinates NbFd, which leads to the degradation of NbFd via the ubiquitin-proteasome system (Fig. 4b-d). There are seven lysine residues distributed in NbFd, with four in the signal peptide and three in the C-terminus. However, the exact lysine site(s) in NbFd for the ubiquitination needs further investigation.
Our data suggest that the C-terminal end of XopL is required since XopL-LRR failed to catalyze the degradation of NbFd (Fig. 4c). In the degradation assays, multiple immuno-reactive XopL and XopL-LRR bands were observed at later time points (Fig. 4b, c), suggesting that XopL is modified at the N-terminus when transiently expressed in Nb. The potential modification is ubiquitination but not auto-ubiquitination of the effector, because the E3 ligase deleted mutant (Xop-LRR) of XopL was also modified (Fig. 4c). Although autoubiquitination has been reported for multiple effectors in the IpaH family (Rohde et al. 2007), including NopM in Sinorhizobium fredii (Xu et al. 2018), the effort to show GST:XOPL self-ubiquitination in vitro was unsuccessful (Additional file 7: Figure S5), which indicates that XopL might not be able to ubiquitinate itself.
Stromules (stroma-filled tubules that emanate from plastid bodies) are highly abundant during ETI in Nb and are potentially involved in the transport of defense signals from the chloroplast to the nucleus (Caplan et al. 2015). Recently, it was reported the expression of XopL Xcv in Nb resulted in massive changes in cell morphology including cells that were almost completely devoid of stromules. In contrast, a mutant of XopL Xcv containing the LRR domain but lacking E3 ligase activity failed to eliminate stromules and was localized to microtubules (Erickson et al. 2018). The authors speculated that the XopL mutant likely binds potential XopL targets through the LRR domain, but is unable to ubiquitinate them. In our work, we observed that XopL-LRR interacted with NbFd in the yeast twohybrid and BiFC assays (Fig. 3); this supports a model in which the LRR domain of XopL interacts with NbFd. Additionally, Fd2, a foliar ferredoxin in Arabidopsis, was found to be localized to stromules when transiently expressed in Nb. The Fd2 knock-out mutant showed compromised defense and reduced ROS accumulation, indicating a role for Fd2 in the plant immune response . The authors suggested that Fd2 might function in retrograde signal transduction from the chloroplast to the nuclear envelope via stromules. Although stromules were not evaluated in our study, it remains plausible that the XopL-mediated degradation of NbFd may cause changes of stromules.
Singer et al. hypothesize that XopL has multiple roles including the suppression of PTI via its LRR region, the ubiquitination of a plant substrate that may be recognized by the LRR region, and the elicitation of programmed cell death that requires both the LRR region and XL-box of XopL (Singer et al. 2013). Our yeast twohybrid results indicate that NbFd interacts with both the LRR-and XL-box-containing domains of XopL Xoo, but these two separate domains have lower binding abilities than the full length XopL towards NbFd (Fig. 3b). Thus we speculate that NbFd binds to XopL at the junction of LRR and XL-box with two binding sites. The interaction of XopL with NbFd and subsequent ubiquitination of NbFd is an interesting and novel finding, but the evolutionary forces resulting in the ubiquitination of NbFd by a Xanthomonas effector remain a mystery. Although ubiquitin and ferredoxin have no obvious analogy in function or conserved sequence motifs, it is important to note that structural similarities between these two proteins have been reported previously (Michnick and Shakhnovich 1998). Ferredoxin has topological features and dimensions of structural elements that are conserved in ubiquitin, and the two proteins share a conserved folding motif known as the β-grasp (Vriend and Sander 1991;Fukuyama 2004). It is interesting to speculate that ubiquitin and ferredoxin may have a common ancestor and perhaps their shared topology might help explain how XopL evolved to recognize and interact with these two proteins.

Conclusions
In the study we report the cell death-inducing ability of four type III secretion system effectors of X. oryzae in Nb. Specially, one of the effectors, XopL, holding LRR region at its N-terminus and E3 ubiquitin ligase domain at the C-terminus, interacts with Nb ferredoxin (NbFd).
The interaction results in the ubiquitination and degradation of NbFd and subsequently leads to ROS production and defense-related response in plants. Taken together, our findings enlarge the repertoire of effector targets to include key components of plant defense systems.

Bacterial strains, plasmid and plant materials
The bacterial strains and plasmids used in this study are listed in Additional file 4: Table S2. Escherichia coli strains were cultivated in Luria-Bertani (LB) medium at 37°C (Sambrook and Russell 2001). X. oryzae pv. oryzae PXO99 A and X. oryzae pv. oryzicola RS105 were cultivated in nutrient broth (NB) or NB with agar at 28°C (Li et al. 2011). Agrobacterium was grown in LB medium containing rifampicin at 28°C. Saccharomyces cerevisiae was grown in Yeast Extract Peptone Dextrose medium at 28°C. Antibiotics were used at the following final concentrations: ampicillin, 100 μg/mL; rifampicin, 75 μg/mL; and kanamycin, 25 μg/mL.
Nb plants were cultivated at 25°C in a growth chamber with a 16 h light/8 h dark photoperiod.

Detection of cell death
Six-week old leaves were inoculated for transient expression of effectors as described above. At 7 dpi, the inoculated leaves were collected and washed in water, then submerged in a 0.25% (w/v) aqueous solution of Evans blue for 24 h at room temperature. The stained leaves were washed three times (15 min/wash) with gentle agitation in PBS plus Tween (0.05%, v/v) to remove excess dye (Wright et al. 2000). Leaves were then transferred to 95% ethanol in a boiling water bath for bleaching until chlorophyll was completely removed. Samples were stored and examined in 65% ethanol and photographed under white light. Three independent plants were used, and experiments were repeated three times with similar results.

DAB staining for ROS accumulation
In situ H 2 O 2 accumulation was detected using 3, 3′-diaminobenzidine (DAB) staining as described previously (Bindschedler et al. 2006) with minor modifications. Briefly, Agrobacterium-infected six-week old Nb leaves were collected at different time points and submerged in a DAB (1 mg/mL) solution containing 0.05% Tween-20 (v/v) and 10 mM sodium phosphate buffer (pH 7.0) under darkness for 8 h. Leaves were carefully transferred to 95% ethanol and placed in a boiling water bath until chlorophyll was completely removed. Samples were stored in 65% ethanol and then photographed under white light. N-acetyl-L-cysteine (NAC; 2 mM) was used as a ROS scavenger and co-infiltrated with Agrobacterium. Three independent plants were used, and experiments were repeated three times.

Yeast two-hybrid assays
The Matchmaker™ GAL4 two-hybrid system (Clontech, CA) was used to screen for XopL-interacting proteins in this study. The xopL coding sequence in pGBKT7-xopL was used to generate bait for screening a Nb cDNA library of mature leaves. SD medium was used for screening assays. 1× Amino Acid Dropout Mix (Clontech, CA) was added to SD, including SD/−Trp-Leu-His (lacking tryptophan, leucine, and histidine) and SD/ −Trp-Leu-His-Ade (lacking tryptophan, leucine, histidine and adenine).
The yeast strain AH109 was co-transformed with pGBKT7-xopL and the Nb cDNA library and cultured on SD/−Leu-Trp-His medium. Transformants were then transferred to SD/−Leu-Trp-His-Ade agar medium containing 20 mg/mL X-α-GAL. Blue colonies were considered to be potential positive interactions; positives were amplified using primer pair T7/3AD (Additional file 8: Table S3), and PCR products were sequenced.
Interactions of XopL, NbFd and their derivatives were analyzed on SD/−Leu-Trp-His-Ade/X-α-GAL medium. Briefly, transformants were grown on SD/−Leu-Trp-His medium and then transferred to SD/−Leu-Trp-His-Ade agar medium. Single colonies from SD/−Leu-Trp-His-Ade agar medium were picked and inoculated into 1 mL SD/−Leu-Trp-His-Ade liquid medium with shaking for 24 h. Yeast cells were harvested by centrifugation (5000 rpm, 2 min) and resuspended in sterilized deionized water (OD 600 = 0.5). The water-suspended yeast was 10fold serial diluted and 10 μL was spotted on SD/−Leu-Trp-His-Ade agar medium with and without 20 mg/mL X-α-GAL.

Localization and bimolecular fluorescence complementation (BiFC) assays
BiFC assays were performed as described previously (Walter et al. 2004) with minor modifications. Buffersupplemented Agrobacterium strains containing pXopL-YN and pNbFd-YC were mixed (1:1 ratio) and incubated at 25°C for 1 h. Controls included cotransformation with XopL:YN and empty YC vector and NbFd:YC (Fd:YC) with empty YN vector. Fluorescence from infiltrated Nb leaves was monitored and imaged at 48 hpi with a confocal laser fluorescence microscope Leica TCS SP5-II Confocal Microscope (Leica Microsystems, Wetzlar, Germany). Chloroplasts were visualized by auto-fluorescence.

In vitro ubiquitination assay
To evaluate the potential ubiquitination of NbFd by XopL, we used the Ubiquitin Conjugation Kit K-960 (Boston Biochem, Cambridge, MA, USA). Ubiquitin aldehyde (Boston Biochem) was used to inhibit deubiquitinating enzymes. 10 μL conjugation fraction A, 5 μM GST:XopL, 1.7 μL ubiquitin solution and 1 μL ubiquitin aldehyde were gently mixed and incubated for 5 min at 37°C. Next, 2.5 μL energy solution and 2 μg NbFd:His 6 were added, gently mixed, and incubated for 4 h at 37°C. The reaction was terminated with 10 mM DTT; 5× loading buffer was added and the reactions were boiled for 8 min. The samples were separated by 10% SDS-PAGE, transferred to PVDF membranes (0.2 μm) and probed with anti-His antibody (Abgent, San Diego, USA). After washing with stripping buffer (glycine 1.88 g/L, SDS 20 g/L, pH 2.0), the membrane was re-probed with anti-GST antibody (TransGen Biotech, China). After a final stripping step, the membrane was probed a third time using ubiquitin antibody (Abmart, China).

NbFd degradation assay
To detect the potential degradation of NbFd by XopL and XopL-LRR in planta, pHB-xopL or pHB-XopL-LRR were co-infiltrated with pNbFd-YFP into Nb leaves at a 3:2 ratio. At 2, 3, and 4 dpi, total proteins were extracted with lysis buffer (described above in the CoIP assay). Protein concentrations were determined using the Bradford Coomassie blue method and adjusted to be identical. Total proteins (100 μL) were mixed with 5× SDS loading buffer and boiled for 8 min. Proteins were separated by 10% SDS-PAGE and then transferred to PVDF membranes (0.2 μm). For initial immunoblotting, an anti-myc antibody was used to detect NbFd:YFP and visualized by chemiluminescence. The antibody complex was then removed with stripping buffer, blocked and probed with anti-flag antisera to detect XopL. The membrane was stripped a final time and incubated with anti-actin antibody to detect the total protein level. The role of the ubiquitin proteasome system in degradation of NbFd was investigated by introducing proteasome inhibitor MG132 (100 μM) into the agroinfiltrated leaves 16 h before samples were harvested.

Supplemental methods
See Additional file 9.
isolation and gene expression analysis. Method S5, ion leakage test. Methods S6, overexpression and RNAi-mediated silencing of NbFd.