The E3 Ligase APIP10 Connects the Effector AvrPiz-t to the NLR Receptor Piz-t in Rice

Although nucleotide-binding domain, leucine-rich repeat (NLR) proteins are the major immune receptors in plants, the mechanism that controls their activation and immune signaling remains elusive. Here, we report that the avirulence effector AvrPiz-t from Magnaporthe oryzae targets the rice E3 ligase APIP10 for degradation, but that APIP10, in return, ubiquitinates AvrPiz-t and thereby causes its degradation. Silencing of APIP10 in the non-Piz-t background compromises the basal defense against M. oryzae. Conversely, silencing of APIP10 in the Piz-t background causes cell death, significant accumulation of Piz-t, and enhanced resistance to M. oryzae, suggesting that APIP10 is a negative regulator of Piz-t. We show that APIP10 promotes degradation of Piz-t via the 26S proteasome system. Furthermore, we demonstrate that AvrPiz-t stabilizes Piz-t during M. oryzae infection. Together, our results show that APIP10 is a novel E3 ligase that functionally connects the fungal effector AvrPiz-t to its NLR receptor Piz-t in rice.


Introduction
Unlike animal responses to pathogen infection, plant responses to pathogen infection do not include a circulatory system or specialized cells [1]. Instead, individual plant cells launch defense responses against invading pathogens. Extensive molecular studies over the last two decades have revealed two layers of host immunity in plants. Plant immunity can be activated when highly conserved pathogen-associated molecular patterns (PAMPs) are recognized by plasma membrane-bound pattern recognition receptors (PRRs) in a process called PAMP-triggered immunity (PTI). PTI is considered the first layer of plant immunity [2,3]. For the second layer, immunity can be activated when pathogen-delivered avirulence (Avr) effectors are recognized by the product of plant resistance (R) genes in a process called effector-triggered immunity (ETI). ETI can be achieved by the direct or indirect interaction between the Avr effectors and R proteins in the plant cell [1,4]. Upon recognition, both immunity layers are capable of initiating a signaling cascade that can result in multiple defense responses.
The nucleotide-binding domain, leucine-rich repeat (NLR) proteins play a major role as intracellular immune receptor R proteins in plant immunity [5]. Most R genes cloned to date encode NLR proteins that mediate recognition of diverse effectors from all classes of plant pathogens. Both direct and indirect interactions between NLRs and effectors occur in different pathosystems [6]. In the indirect interactions, additional plant proteins are the targets of effectors and may be either genuine virulence targets of the effectors [7] or decoy proteins that plants have evolved to mimic bona fide effector targets [8]. A hybrid model of the direct and indirect interactions was proposed in which the target protein serves as 'bait' that the effector associates with before direct interaction with the NLR receptor and before immune signaling is activated [9,10].
Because NLR activation and signaling usually results in strong defense responses and a hypersensitive reaction (HR), such activation and signaling must be tightly regulated to avoid adverse effects on plant growth and development when plants are not under pathogen attack [11][12][13]. Some factors controlling NLR activation and signaling have been identified [10]. For example, analysis of crystal structure and in vitro interactions revealed that the two CHORD domains of a single RAR1 molecule bridge the N-termini of the HSP90 monomers, thus regulating the 'open' and 'closed' state of the HSP90 dimer that coordinates NLR stabilization [14]. Two independent studies showed that the tetratricopeptide repeat-containing protein SRFR1 is a negative regulator of the accumulation and activation of the NLR receptor SNC1 [15,16].
Ubiquitin-mediated degradation of proteins via the 26S proteasome is important for the regulation of protein levels in living cells [17]. The E3 ligases in the ubiquitination process interact and bring substrates to be ubiquitinated in proximity to the conjugating enzyme E2. Involvement of ubiquitination in NLR-mediated immunity has been recently reported in plants (see review by [18]). For example, the SCF E3 ubiquitin ligase complex involving the Fbox protein CPR30/CPR1 (SCF CPR1 ) targets the NLR protein SNC1 and RPS2 for degradation [11,12]. Furthermore, when plants overexpressing CPR1 were treated with the 26S proteasomal inhibitor MG132, levels of SNC1 increased. This observation suggested that SNC1 levels are modulated by SCF CPR1 via the 26S proteasome-mediated degradation pathway. However, it is unknown whether these E3 ligase complexes are the host targets of pathogen effectors.
Magnaporthe oryzae is the causal agent of rice blast, a severe disease that limits rice production worldwide. We cloned the R gene Piz-t from rice and its cognate avirulence gene AvrPiz-t from M. oryzae [4,19]. Piz-t encodes an NLR protein, but AvrPiz-t does not show homology to any genes in the databases. Our previous study showed that AvrPiz-t is secreted into the rice cell and interacts with the rice E3 ligase APIP6 [20]. Analysis of APIP6 RNAi transgenic plants showed that APIP6 is a positive regulator of PTI. AvrPiz-t and APIP6 can degrade each other in planta. APIP10 is another E3 ligase that was identified in the yeast-two hybrid screen when AvrPiz-t was used as the bait in our previous study [20]. To determine the role of APIP10 in rice immunity in the current study, we analyzed the relationships among APIP10, AvrPiz-t, and Piz-t. We report that APIP10 contributes to rice immunity by functioning in both PTI and ETI. While APIP10 is the target of AvrPiz-t's virulence activity for suppression of PTI, APIP10 is also a negative regulator of the accumulation and activation of the NLR receptor Piz-t. Degradation of APIP10 by AvrPiz-t results in the release of APIP10 suppression of Piz-t accumulation and the initiation of a strong defense response. Our study identifies a novel E3 ligase, APIP10, that functionally connects a fungal effector with its NLR receptor in plants.

AvrPiz-t interacts with the putative E3 ligase APIP10 in vitro and in vivo
We previously reported that AvrPiz-t interacts with three putative C3HC4-type E3 ligases in yeast-two hybrid (Y2H) screens [20]. Among the AvrPiz-t interacting proteins (APIPs), APIP10 showed a strong interaction and was selected for further studies. We confirmed the interaction between AvrPiz-t and APIP10 in yeast using four selection markers (S1A Fig). To further validate the interaction in yeast, we performed in vivo Co-IP (co-immunoprecipitation) experiments by co-expressing GFP:AvrPiz-t:HA and Myc:APIP10 in Nicotiana benthamiana using the agroinfiltration method. Because the AvrPiz-t signal peptide is cleaved in the mature protein, we used the AvrPiz-t gene without its N-terminal signal peptide sequence to make constructs in the following experiments unless indicated otherwise. When the GFP:AvrPiz-t:HA fusion protein was immunoprecipitated (IP) from the plant extract using anti-HA IgG beads, the Myc:APIP10 proteins were detected in the immunocomplex of GFP:AvrPiz-t:HA with the anti-Myc antibody (S1B Fig, middle  Since AvrPiz-t interacts with three putative C3HC4-type E3 ligases in the Y2H screens as previously reported [20], we decided to confirm whether that the interaction between AvrPiz-t and APIP10 is specific by including an unrelated E3 ligase, SPL11 [21], as an additional negative control in the Co-IP experiment. To facilitate the detection of SPL11 in planta as reported previously [22], mSPL11, which has three mutations in the U-box domain, was used in the experiment. Similar with the assay in S1B Fig, GFP:AvrPiz-t:HA, Myc:APIP10 and Myc: mSPL11 were co-expressed in N. benthamiana. When the GFP:AvrPiz-t:HA fusion protein was immunoprecipitated from the plant extract using the anti-HA IgG beads, the Myc:APIP10 proteins were also detected in the immunocomplex of GFP:AvrPiz-t:HA with the anti-Myc antibody (S1C Fig, first  These results indicate that the interaction in planta between AvrPiz-t and APIP10 is specific.
APIP10 is a functional RING finger E3 ligase APIP10 identified from the Y2H library consists of a 1,416-bp open reading frame and encodes a protein with 472 amino acids. BLAST searches showed that APIP10 is a single-copy gene in the rice genome and that the predicted protein belongs to a large family of conserved and ubiquitous RING finger proteins in eukaryotes. APIP10 has multiple, predicted conserved domains such as a BRAP2 (BRCA1-Associated Protein 2), a C3HC4-type RING finger, a ZnF UBP To determine whether APIP10 has E3 ligase activity like other BRAP2-RING Finger-ZnF UBP proteins, we performed an in vitro ubiquitination assay with a series of negative controls. In the E3 ligase activity assay, the MBP:APIP10 fusion protein purified from E. coli was incubated with wheat (Triticum aestivum) E1, Arabidopsis E2 (AtUBC10), 5X Myc:ubiquitin (Myc: Ub), and ATP. To verify whether the E3 ligase activity of APIP10 depends on its RING finger domain, we included the mutant dRING (APIP10 dRING) in which the RING finger at the amino acid position from 158 to 204 was completely deleted. In the preliminary experiment, no polyubiquitin product was visible when Ub was used. Therefore, Ub fused with 5X Myc was used in the reaction, and the immunoblot with anti-Myc antibody was performed to enhance the detection of the polyubiquitin. A band of polyubiquitin product with high molecular weight was detected in the reaction with the complete APIP10 (S2B Fig, lane 6

AvrPiz-t interferes with the E3 ligase activity of APIP10 and is ubiquitinated by APIP10 in vitro
To assess whether AvrPiz-t has any biochemical functions in the APIP10-mediated ubiquitination, we included the purified GST:AvrPiz-t:HA protein in the in vitro E3 ligase assay of APIP10. Surprisingly, a band above the GST:AvrPiz-t:HA protein was detected by the immunoblot analysis with the anti-HA antibody in the presence of the GST:AvrPiz-t:HA recombinant protein (Fig 1A, lane 6, arrow). In contrast, no such signal was detected in the presence of the GST:AvrPi-ta:HA protein, an unrelated effector protein from M. oryzae [23] (Fig 1A, lane  7). The difference in size between the two bands is about the size of a 5X Myc:ubiquitin (about 16 KDa), indicating that the GST:AvrPiz-t:HA is ubiquitinated by APIP10 in vitro.
To confirm the ubiquitination result, we repeated the E3 ligase assay for the reactions in lane 6-8 in Fig 1A. Then we conducted a GST pulldown analysis by washing the GST:AvrPiz-t: HA and GST:AvrPi-ta:HA beads with the PBST solution (Phosphate Buffered Saline with 0.5% Triton X-100) after the E3 ligase reaction. Immunoblot analysis with the anti-HA antibody showed the ubiquitinated bands above the GST:AvrPiz-t:HA fusion protein (S4 Fig, first  Furthermore, the immunoblot analysis that used the anti-Myc antibody to detect polyubiquitin showed that the E3 ligase activity of APIP10 was reduced when the GST:AvrPiz-t:HA fusion protein was included in the reaction (Fig 1B, lane 6, arrow), suggesting that AvrPiz-t interferes with APIP10 E3 ligase activity. The GST:AvrPi-ta:HA protein, which did not affect the E3 ligase activity of APIP6 [20], was used as a negative control (Fig 1B, lane 7). These results show that AvrPiz-t interferes with APIP10 E3 ligase activity and that APIP10 ubiquitinates AvrPiz-t in vitro.
APIP10 and AvrPiz-t promote degradation of each other through the 26S proteasome system in N. benthamiana Because AvrPiz-t is ubiquitinated by APIP10 in vitro (Fig 1A), we hypothesized that AvrPiz-t may be a substrate of APIP10 in planta. To test this, we co-expressed GFP:AvrPiz-t:HA with Myc:APIP10 in N. benthamiana leaves. The immunoblot analysis revealed that the GFP:Avr-Piz-t:HA protein level was significantly lower in the tissue when Myc:APIP10 was co-expressed than when Myc:APIP10 dRING was co-expressed ( , suggesting that APIP10 degrades AvrPiz-t via the 26S proteasome system. To test whether the mutated APIP10 dRING affects the stability of its associated protein AvrPiz-t, we used Myc:GFP as a negative control and repeated the co-infiltration assay. The western blot analysis revealed that GFP:AvrPiz-t:HA was significantly lower in the tissue  Because AvrPiz-t promotes the degradation of APIP6 when they are co-expressed in N. benthamiana [20], we reasoned that AvrPiz-t might also affect the accumulation of APIP10 in plant cells. The immunoblot analysis showed that APIP10 was reduced by 30-40% (relative to the controls) when GFP:AvrPiz-t:HA and Myc:APIP10 were co-expressed ( Fig 2B, lane 3 in the upper panel). Furthermore, the degradation was inhibited by MG132 ( Fig 2B, lane 4 in the upper panel), suggesting that AvrPiz-t promotes the degradation of APIP10 in planta, likely through the 26S proteasome system. To obtain more evidence that the degradation of APIP10 is dependent of AvrPiz-t, AvrPii, an unrelated effector protein [24], was used as a negative control in the co-infiltration. The immunoblot analysis showed that the accumulation of Myc: Knockdown of APIP10 in the NPB background (Piz-t -) compromises PTI As described above, AvrPiz-t interferes with the E3 ligase activity of APIP10 in vitro ( Fig 1B) and promotes the degradation of APIP10 in vivo ( Fig 2B). Based on these results and our previous report on APIP6 [20], we speculated that APIP10 might be another target of AvrPiz-tmediated suppression of host defense. To determine the role of APIP10 in PTI against M. oryzae, we designed an RNAi construct targeting 215 bp of APIP10 3'UTR using the pCXUN vector [25] and generated over 20 stable transgenic lines in NPB (without Piz-t). Three homozygous lines with a single T-DNA insertion were selected from the T 3 generation for the subsequent analysis and transcription levels of each line were determined by qRT-PCR (S7 Fig,  top panel). To determine the function of APIP10 in rice PTI, we used a luminol-based chemiluminescence assay to monitor ROS generation induced by flg22 and chitin treatments in 4-week-old homozygous APIP10 RNAi plants [26]. The analysis showed that APIP10 knockdown significantly suppressed flg22-induced ROS accumulation ( Fig 3A) but only partially suppressed chitin-induced ROS accumulation in the rice tissue (Fig 3B). At 3 h after treatment with flg22 and chitin, qRT-PCR also revealed that the transcriptional profiles of PTI-related defense genes such as KS4 and PAL differed between APIP10 RNAi plants and non-silenced plants ( Fig 3C). KS4 encodes one of the two diterpene cyclase enzymes involved in momilactone biosynthesis [27,28], and the APIP10 RNAi plants showed significant suppression of KS4 transcripts 3 h after treatment with flg22 ( Fig 3C, left). Phenylalanine ammonia lyase (PAL) catalyzes the deamination of L-phenylalanine to trans-cinnamic acid and is involved in the biosynthesis of certain classes of low molecular weight antimicrobial compounds called phytoalexins [29,30]. In rice, the transcription of PAL is induced by flagellin from bacteria as well as by chitin [31][32][33]. In APIP10 RNAi lines, significant suppression of PAL transcripts was observed at 3 h after treatment with either flg22 or chitin (Fig 3C, right). Suppression of PAL transcripts without PAMP treatment was also observed (Fig 3C, water), indicating that APIP10 might be transcriptional level of each gene was determined by semi-quantitative (sq)-PCR. dRING denotes APIP10 dRING. The experiments were performed at least three times with similar results. (B) Degradation of APIP10 when it is co-expressed with the AvrPiz-t protein in N. benthamiana. Tissues were harvested 3 days after infiltration. MG132 (50 μM) was infiltrated with DMSO as a control at 18 h before sampling. The experiments were performed at least three times with similar results.  involved in PAMP-independent regulation of PAL. Taken together, these results suggest that host immunity triggered by PAMPs is compromised in APIP10 RNAi plants.
Next, we used the punch inoculation method to measure the resistance of APIP10 RNAi plants to the virulent M. oryzae isolate RB22. Lesions were larger on the APIP10 RNAi plants than on the control plants ( Fig 3D). The relative fungal mass of M. oryzae, measured by the DNA-based qPCR assay, was also greater in the APIP10 RNAi plants than in the control plants ( Fig 3E, left panel). In addition, more spores were produced on the APIP10 RNAi plants than on the control plants ( Fig 3E, right panel). These results suggest that silencing of the APIP10 gene in rice compromises the basal defense against M. oryzae. Because the phenotypes of APIP10 RNAi lines are similar to those of APIP6 RNAi lines, we conducted qRT-PCR of APIP6 in the APIP10 RNAi lines (. S8A Fig, top panel) and APIP10 in the APIP6 lines (S8B Fig,  lower panel). The analysis showed that there is no correlation between the expression levels of the two genes when one of them is silenced.

Knockdown of APIP10 in the Piz-t background causes severe spontaneous cell death phenotypes
Piz-t is the cognate R gene of AvrPiz-t in rice [4]. To understand the relationship between APIP10 and Piz-t, we first transformed the Piz-t:HA construct with hygromycin as a selection marker in the NPB background in order to generate NPB Piz-t:HA rice. Then we screened T1 lines for hygromycin resistance and chose the lines showing a 3:1 segregation ratio. After the T 2 plants were inoculated with the avirulent isolate RO1-1, the presence of the Piz-t:HA fusion protein was confirmed by immunoblot analysis with the anti-HA antibody. With this approach, we obtained five homozygous lines with a single T-DNA insertion of Piz-t:HA. Next, we developed an APIP10 RNAi construct (same as described above) with the G418 Sulfate antibiotic selection and transformed the construct into the calli of selected NPB Piz-t:HA line. The APIP10 transcription levels in selected lines were determined by qRT-PCR (S7 Fig, lower  panel). Surprisingly, NPB Piz-t:HA calli transformed with the APIP10 RNAi construct showed severe cell death but not with an empty vector ( Fig 4A and S9 Fig).
To determine whether the severe cell death triggered in the calli is related to the Piz-t:HA protein, we performed immunoblot analysis for Piz-t:HA with the anti-HA antibody and qRT-PCR of the APIP10 transcripts; we did this with four independently transformed callus lines that showed different levels of cell death. qRT-PCR indicated that the transcript level of APIP10 was lower in lines with severe cell death than in the line without cell death (S10 Fig, the  upper panel), indicating that the cell death observed in the transgenic plants was caused by the silencing of APIP10. Strikingly, the lines with lower APIP10 transcripts in the NPB Piz-t:HA background showed a higher expression level of the Piz-t-HA protein ( Fig 4B, the upper  panel), suggesting that the APIP10 transcript level is negatively correlated with Piz-t accumulation and cell death. To determine whether the accumulation of the Piz-t protein is due to its increased transcription, we used qRT-PCR to quantify Piz-t transcripts. The analysis revealed that the number of Piz-t transcripts in the APIP10 RNAi callus lines was inversely related to Piz-t protein accumulation (.S10 Fig, the lower panel), indicating that the accumulation of the Piz-t protein is not the result of its increased transcription.
After two rounds of rice transformation, we were able to obtain only six T1 lines that had reduced cell death because most APIP10 RNAi lines died within 2-3 weeks after transfer to soil ( Fig 4C, right panel). To confirm the results obtained from callus lines, we compared the T 2 plants of the APIP10 RNAi lines in the NPB Piz-t:HA background with the wild-type NPB Pizt:HA plants. All of the lines in the NPB Piz-t:HA background showed cell death and dwarf phenotypes (S11A Fig). Immunoblot analysis of total protein obtained from the leaf tissue of two lines (#14-4 and 38-6) revealed that they contained 1.5-to 2.2-times more Piz-t:HA than the controls (Fig 4D). qRT-PCR analysis using the same leaf tissue showed that Piz-t transcripts were less abundant in the APIP10 RNAi lines 14-4 and 38-6 than in the control plants (S11B Fig), indicating that the Piz-t accumulation is not due to an increase in its transcription level.
Knockdown of APIP10 in the Piz-t background compromises PTI but still enhances resistance to a virulent isolate To determine whether PTI responses of APIP10 RNAi lines in NPB Piz-t:HA were compromised as we had observed in the APIP10 RNAi lines in NPB, we monitored the ROS generation triggered by chitin and flg22 in the leaf disks of 4-week-old plants. The analysis showed that silencing of APIP10 in the presence of Piz-t almost completely suppressed ROS induced by flg22 in the leaf tissue (Fig 3F) but did not suppress ROS induced by chitin significantly ( Fig  3G). This observation was similar to that observed in the APIP10 silencing lines in NPB ( Fig  3B). qRT-PCR also revealed that the RNAi lines showed significant but distinct transcriptional induction or suppression profiles of the PTI-related defense genes 3 h after treatment with chitin or flg22 (Fig 3H). For example, the APIP10 RNAi lines showed suppression of KS4 transcripts at 3 h after treatment with flg22 or chitin (Fig 3H, left) but showed significant suppression of PAL transcripts at 3 h after treatment with flg22 but not after treatment with chitin (Fig 3H, right).
We next used the punch inoculation method to measure the resistance of the APIP10 RNAi lines in the NPB Piz-t:HA background to the virulent M. oryzae isolate RB22. Surprisingly, lesions were smaller on the APIP10 RNAi plants in the Piz-t background than on the wild-type NPB-Piz-t:HA plants ( Fig 3I); this was opposite to the disease phenotype observed in the APIP10 RNAi plants in the NPB background (Fig 3D). Consistent with lesion size, the relative fungal biomass measured by the DNA-based qPCR was significantly lower in the APIP10 RNAi-NPB-Piz-t:HA plants than in the Piz-t:HA plants (Fig 3J, left panel). Furthermore, the sporulation in the infected area was significantly lower on the APIP10 RNAi-NPB-Piz-t:HA plants than on the Piz-t:HA plants (Fig 3J, right panel). Taken together, these results suggest that silencing of APIP10 in the Piz-t plants enhances resistance to the virulent isolate RB22 possibly through the accumulation of Piz-t, even though the PTI response was compromised because of the silencing of APIP10.

APIP10 promotes degradation of the Piz-t protein through the 26S proteasome system
Because we found that silencing of APIP10 in transgenic rice leads to accumulation of the Piz-t protein (Fig 4B and 4D), we speculated that APIP10 may promote the degradation of Piz-t in rice. To test this hypothesis, we co-expressed either Myc:APIP10 or Myc:APIP10 dRING (as a negative control) with Piz-t:HA using agroinfiltration in N. benthamiana leaves. Less Piz-t:HA protein accumulated when it was co-expressed with APIP10 than when it was co-expressed with APIP10 dRING, and the accumulation was recovered almost to the control level by treatment with MG132 (S12 Fig, lane 1

vs. 2 and lane 3 vs. 4 in the first panel).
To confirm these results, a semi-in vivo degradation assay was performed as described previously [34,35]. For this assay, total protein extracted from the M. oryzae-inoculated Piz-t:HA rice plants was mixed with total protein extracted from N. benthamiana leaves co-expressed with either APIP10 or APIP10 dRING with GFP as an internal control. The samples were collected at different times before 1X SDS loading buffer was added to stop the reaction. The level of Piz-t was decreased by APIP10 over time, and after 2 h, less Piz-t protein was detected in the presence of APIP10 (Fig 5A, lane 5) than in the presence of APIP10 dRING (Fig 5A,  lane 10). The decrease in the level of the Piz-t protein at 4 h was partially inhibited by treatment with MG132 ( Fig 5B, lane 2 in the first panel). To rule out the possibility that the deletion of the RING finger domain in APIP10 stabilizes both APIP10 dRING and Piz-t, myc:GFP was used as a negative control instead of APIP10 dRING in the semi in vivo degradation assay. The decrease in the level of the Piz-t protein was observed in the presence of Myc: APIP10 (S13 Fig, lane 1 vs. lane 3 in the top panel) and the degradation of Piz-t:HA was partially inhibited by the MG132 treatment (S13 Fig, lane 1 vs. lane 2). However, no obvious degradation of Piz-t in the presence of Myc:GFP was observed (S13 Fig, lane 3 vs. lane 4 in the top panel). Together, these data suggest that APIP10 promotes Piz-t degradation through the 26S proteasome system. Degradation of Piz-t by APIP10 depends on APIP10 E3 ligase activity. Total protein from the inoculated rice plants was mixed with total protein from N. benthamiana in which either Myc:APIP10 or Myc:APIP10 dRING (dRING) was co-expressed with GFP. PBZ1 or GFP was used as a loading control for Piz-t and Myc:APIP10 or Myc:dRING, respectively. (B) Suppression of Piz-t degradation by MG132 treatment. As in Fig 2A, MG132 was added to inhibit the degradation of proteins through the 26S proteasome system; the reaction was stopped at 4 h after incubation. PR1 or GFP was used as a loading control for Piz-t:HA and Myc:APIP10 or Myc:APIP10 dRING (dRING), respectively.

AvrPiz-t promotes the accumulation of the Piz-t protein in planta
Because we found that AvrPiz-t promotes degradation of APIP10 and that silencing of APIP10 leads to accumulation of Piz-t, we reasoned that expression of AvrPiz-t in planta may lead to the accumulation of Piz-t. To determine the relationship between AvrPiz-t and Piz-t, we coexpressed either empty vector, GFP, or GFP:AvrPiz-t with Piz-t:HA in N. benthamiana and observed the accumulation of both proteins by immunoblot analysis. Intriguingly, the Piz-t protein level was~1.6-fold greater when Piz-t:HA was co-expressed with GFP:AvrPiz-t:HA (Fig 6A, top panel, lane 3) than with the empty vector or with GFP ( Fig 6A, top panel, lane 1  and 2), suggesting that AvrPiz-t stabilizes the Piz-t protein.
To determine whether the increased accumulation of Piz-t in the presence of AvrPiz-t is specific to AvrPiz-t, we included an unrelated effector protein from M. oryzae, AvrPii, as a negative control. As described above, we co-expressed the empty pGD vector, GFP:AvrPiz-t or GFP:AvrPii with Piz-t:HA in N. benthamiana and observed the accumulation of proteins by immunoblot analysis. The assay showed that the Piz-t protein level was~2.7-fold greater when Piz-t:HA was co-expressed with GFP:AvrPiz-t:HA (S14 Fig, the  To confirm this result in rice plants, we transformed the construct expressing the AvrPiz-t gene under the control of the inducible XVE system into the rice calli generated from the Piz-t: HA rice seeds in order to generate iAvrPiz-t transgenic lines. Over 20 independently transformed transgenic lines were obtained. After confirming the genotype by PCR, we selected six T 4 lines with the Piz-t plants as a negative control for the following analyses. Young leaves of each line were cut and transferred to N6 medium containing the β-estradiol to induce the expression of AvrPiz-t. Piz-t accumulation increased upon the induction of AvrPiz-t (Fig 6B). The induction of AvrPiz-t as well as the level of Piz-t transcript was monitored by qRT-PCR after the treatment as shown in S15 Fig  To confirm the above results with rice plants under physiologically relevant conditions, i.e., when rice plants were inoculated with M. oryzae, we inoculated the Piz-t plants by the spray method with RB22 isogenic blast transformants: RB22, RB22:AvrPiz-t, and RB22:AvrPii. Consistent with the results from N. benthamiana, Piz-t:HA accumulation was greater in the RB22: AvrPiz-t-inoculated plants than in the RB22 or RB22:AvrPii-inoculated plants at all the time points we monitored (Fig 6C). The Piz-t level was still high in the RB22:AvrPiz-t-inocualted plants at 144 h after inoculation, confirming that AvrPiz-t promotes the accumulation of the Piz-t protein in planta.

Discussion
In host-microbe interactions, ubiquitination plays an important role in both host defense and pathogen infection [18,36]. Researchers have identified both pathogen effectors that target the ubiquitination machinery for defense suppression and plant ubiquitination-related proteins that target pathogen effectors for degradation and thus for defense enhancement. Most of these findings, however, have been derived from several model plant diseases caused by bacterial pathogens. Although recent research has revealed the importance of ubiquitination in plant diseases caused by fungi or oomycetes [20,37], the role of ubiquitination in these diseases is poorly understood. In particular, how microbial perturbation of host ubiquitination activates the defense response mediated by NLR receptor proteins is largely unknown. In this study, we describe the relationship among the fungal effector AvrPiz-t, the rice RING finger E3 ligase APIP10, and the NLR receptor Piz-t. We found that AvrPiz-t can interact with and promote the degradation of APIP10 when it is secreted into rice cells by M. oryzae. Silencing of APIP10 AvrPiz-t. The native promoter in the Piz-t genomic construct was replaced by the 35S promoter to express the Piz-t gene in N. benthamiana. Tissues were harvested at 3 days after agroinfiltration. As negative controls, the vectors pGD and pGDG were expressed. TAP tag was expressed as an internal control to check the efficiency of agroinfiltration. (B) Immunoblot with anti-HA antibody was used to indicate levels of Piz-t after in the Piz-t background causes severe cell death and accumulation of Piz-t. Co-infiltration assays showed that APIP10 expression leads to Piz-t degradation and that such degradation depends on the RING finger domain of APIP10. In contrast, AvrPiz-t can stabilize Piz-t during blast infection. These results demonstrate an elegant defense mechanism in which rice cells use the E3 ligase APIP10 to regulate Piz-t for immune responses when attacked by the blast fungus carrying the AvrPiz-t gene. Degradation of the APIP10 by AvrPiz-t and the resulting elimination of the negative regulation of Piz-t protein by APIP10 leads to a rapid increase in Piz-t and a strong defense response including programmed cell death in the infected cells. Although APIP10 seems similar to RIN4 in Arabidopsis in that they both negatively regulate an NLR [7], the ability of APIP10 to degrade a pathogen effector (AvrPiz-t) and to regulate a host NLR (Piz-t) is unique. Therefore, identification of the novel E3 ligase APIP10 that functionally connects a fungal effector to its cognate NLR receptor in rice provides new insights into plant innate immunity.
We found that AvrPiz-t interacts with APIP10 and that this interaction reduces APIP10 E3 ligase activity. APIP10 specifically ubiquitinates AvrPiz-t but not the unrelated effector AvrPita. The ubiquitination of AvrPiz-t depends on the RING domain of APIP10, and the degradation of AvrPiz-t depends on the 26S proteasome system. These results suggest that AvrPiz-t could be the pseudo-substrate of APIP10. By acting as the pseudo-substrate, AvrPiz-t might prevent APIP10 from ubiquitinating either the host proteins or even other cytoplasmic effector proteins secreted from the fungus. Interaction with AvrPiz-t reduces APIP10's E3 ligase activity and the stability of APIP10. Turnover of E3 ligase enzymes that contain RING domains is regulated in many ways in cells. For example, CBL RING E3 ligases have substrate-dependent regulation in animal systems because substrate specificity plays a role in CBL turnover [38][39][40]. Turnover of the rice E3 ligase APIP10 by a fungal effector is an interesting phenomenon. Further investigation of the mechanism underlying APIP10 degradation by AvrPiz-t in rice may provide new insights into fungal-plant interactions.
Ubiquitination is important in the regulation of defense-related NLR proteins in plants. Recent research shows that SCF CPR1 regulates the levels of SNC1 and RPS2 [11,41]. Also, COP1-dependent regulation of an NLR HRT was observed in response to turnip crinkle virus [13]. In this study, we found that APIP10 is a negative regulator of the NLR receptor Piz-t. Specifically, Piz-t levels are higher in the APIP10 RNAi plants than in control plants, which is consistent with the accumulation of Piz-t after infection with isolates carrying AvrPiz-t. The accumulation of the Piz-t protein after the recognition of AvrPiz-t is unique because the R proteins RPM1 and RPG1 disappear after the recognition of cognate effectors [42,43]. RPM1 is rapidly degraded with effector-mediated activation, perhaps due to a negative feedback that limits HR and attenuates the disease resistance response [42]. The same laboratory reported that the disappearance of activated RPM1 is not dependent on the proteasome system because the two common proteasome inhibitors, MG132 and clasto-lactacystin β-lactone, were not able to stabilize steady-state levels of RPM1 [44]. Our results show that the Piz-t protein level is regulated by the proteasome system. However, we don't know whether the accumulation of the Piz-t protein due to reduced transcription levels of APIP10 leads to the activation of Piz-t and whether the activated form is translocated into the nucleus. In addition, the accumulation of estradiol-mediated induction of AvrPiz-t in NPB Piz-t:HA plants. The leaf samples were collected 24 h after the estradiol treatment. The numbers below the immunoblot indicate relative band intensity. HSP90 was used as a loading control. (C) Immunoblot with anti-HA antibody was used to indicate levels of Piz-t at different time points after inoculation in the NPB Piz-t:HA plants infected with M. oryzae isolate RB22, RB22:AvrPiz-t and RB22:AvrPii. The number below each sample is the relative intensity compared to that of RB22-inoculated sample at each time point. HSP90 was used as a loading control.
In addition, the APIP10 RING domain is vital for Piz-t degradation, suggesting the importance of E3 ligase activity in regulating NLRs. Even though the negative regulation of Piz-t by APIP10 during M. oryzae infection is clear, how APIP10 regulates the Piz-t protein level before and after M. oryzae infection remains unknown. Because no direct interaction between APIP10 and Piz-t was found in yeast, we speculate that a partner or chaperone protein links APIP10 and Piz-t in rice cells. Identifying this unknown protein is one of our objectives in dissecting the Piz-t-mediated signaling pathway.
Many Avr/R protein-protein interactions fit into the guard model, but the molecular mechanism that regulates the recognition and signaling between the pathogen and host proteins is incompletely understood, especially for fungal and oomycete pathogens of plants. In addition and as noted earlier, the regulation of NLR receptors before and after pathogen infection is still unclear. In our previous study [20], we found that AvrPiz-t targets the E3 ligase APIP6 for degradation and that APIP6 in turn degrades AvrPiz-t and positively regulates PTI to M. oryzae in rice. In this study, we found that APIP10 not only has a similar relationship as APIP6 with Avr-Piz-t but also negatively regulates the NLR protein Piz-t via ubiquitination. Based on our previous results and those from this study, we provide the following model to explain the relationships among AvrPiz-t, APIP6, APIP10, and Piz-t (Fig 7). After being secreted and translocated into rice cells, AvrPiz-t interacts with both APIP6/10 and interferes with their E3 ligase activity. In response, APIP6/10 ubiquitinate AvrPiz-t, which causes the degradation of AvrPiz-t in rice cells. When AvrPiz-t is delivered into rice cells without Piz-t, it promotes the degradation of APIP6/10 to suppress PTI. When the Piz-t plants are not attacked by M. oryzae, APIP10 directly or indirectly maintains the Piz-t protein at a low level through ubiquitination. However, when AvrPiz-t is delivered into rice cells expressing Piz-t, AvrPiz-t interacts with and degrades APIP10, which removes the negative regulation on Piz-t and causes the rapid accumulation of Piz-t protein. The rapid accumulation of Piz-t triggers a strong HR and the activation of defense responses.

Materials and Methods
Agroinfiltration assay in N. benthamiana Agrobacterium tumefaciens strain GV3101 carrying different constructs was grown with shaking at 28°C. After 18 h, bacterial cells were spun down for 20 min at 3,200 g and then resuspended in MES buffer (10 mM MgCl 2 and 10 mM MES, pH 5.6) to a final OD 600 of 1.5 for testing constructs, OD 600 of 1.0 for p19, and OD 600 of 0.25 for TAP tag. After acetosyringone was added to a final concentration of 150 μM, bacterial suspensions were kept at room temperature in the dark for 3 h and were then infiltrated into N. benthamiana plants as previously described [45].

Rice plant induction treatment, inoculation, and gene expression analysis
The punch inoculation method [46] with a slight modification was used to evaluate infection of rice plants by M. oryzae. Isolate RB22 was cultured on oat meal agar medium for 2 weeks. A 10-μl volume of a spore suspension (5 × 10 5 spores ml -1 ) was applied to slightly punctured sites of leaves on plants that were 4 to 6 weeks old. Lesion diameter was recorded 10 days after inoculation. To measure the sporulation rate, we removed a 3×1 cm 2 leaf piece that included a lesion and immersed it in a microcentrifuge tube containing 100 μl of distilled water with 1% Tween 20. The samples were vigorously mixed in a vortex apparatus for 2 min to dislodge the spores, and the number of spores ml -1 was determined with a microscope and hemacytometer. The infection ratio was calculated as previously described [47]. Student's ttest was used to test the significance of the differences between the segregating wild-type and lines expressing the transgenes.
qRT-PCR was used to determine gene expression in rice plants or detached leaves. After treatments had been applied to plants or detached leaves, total RNAs were extracted from leaf tissue using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Total RNA was treated with DNase I (Invitrogen) to remove DNA contamination following the manufacturer's protocol. About 1 μg of DNase I-treated RNA was subjected to first-strand cDNA synthesis using the Promega Reverse Transcription System (Promega). qRT-PCR was Fig 7. Working model of the relationships among AvrPiz-t, APIP6, APIP10, and Piz-t. When M. oryzae spores land on a rice leaf and germinate, cellsurface localized PRRs can recognize PAMPs, which leads to PTI as a defense mechanism against the fungal infection. To counteract PTI, M. oryzae secretes effector proteins including AvrPiz-t. Once AvrPiz-t enters rice cells, it is translocated into rice cells and promotes the degradation of APIP6/10, which are positive regulators of PTI, to suppress defense responses such as accumulation of ROS and induction of defense-related genes. However, APIP10 is also a negative regulator of Piz-t in rice cells, and the degradation or modification of APIP10 by AvrPiz-t removes the negative regulation on Piz-t and thus leads to the accumulation of Piz-t, which triggers ETI-mediated responses. To counter its degradation by AvrPiz-t, APIP10 ubiquitinates AvrPiz-t, which leads to the degradation of the effector in the rice cell.
doi:10.1371/journal.ppat.1005529.g007 APIP10 Regulates the NLR Receptor Piz-t carried out using the iQ5 real-time PCR detection system (Bio-Rad). Primers used in this study are listed in S1 Table.

Measurement of ROS level
The second leaf from the top of 4-to 6-week-old rice plants was used for the measurement of ROS. Small leaf discs (approximately 4 mm in diameter) were cut from the leaves with a cork borer and pre-incubated overnight in sterile-distilled water. After the leaf disks were treated with elicitors, ROS generation was monitored by the luminol chemi-luminescence assay [26]. Three pre-incubated leaf disks per sample were immersed in a microcentrifuge tube containing 100 μl of luminol (Bio-Rad), 1 μl of horseradish peroxidase (Jackson ImmunoResearch), and the elicitor (100 nM flg22, 8 nM hexa-N-acetylchitohexaose, or water control). Luminescence was measured at 10-s intervals for 21 min using a Glomax 20/20 luminometer (Promega). Each treatment was represented by three replicate microcentrifuge tubes.

Semi-in vivo degradation of Piz-t:HA by APIP10
Semi-in vivo degradation of Piz-t:HA protein by APIP10 was measured as described before [34]. Briefly, total rice protein extracted from M. oryzae-inoculated Piz-t:HA plants was mixed with total protein from either APIP10-or APIP10 dRING-agroinfiltrated N. benthamiana leaves. The reaction was started by adding 10 μM ATP to the mixture on ice. At the indicated time point, SDS loading buffer was added to the samples, which were boiled for 5 min to stop the reaction before immunoblot was performed.
More detailed experimental methods can be found in S1 Methods.