PARAQUAT TOLERANCE3 Is an E3 Ligase That Switches off Activated Oxidative Response by Targeting Histone-Modifying PROTEIN METHYLTRANSFERASE4b

Oxidative stress is unavoidable for aerobic organisms. When abiotic and biotic stresses are encountered, oxidative damage could occur in cells. To avoid this damage, defense mechanisms must be timely and efficiently modulated. While the response to oxidative stress has been extensively studied in plants, little is known about how the activated response is switched off when oxidative stress is diminished. By studying Arabidopsis mutant paraquat tolerance3, we identified the genetic locus PARAQUAT TOLERANCE3 (PQT3) as a major negative regulator of oxidative stress tolerance. PQT3, encoding an E3 ubiquitin ligase, is rapidly down-regulated by oxidative stress. PQT3 has E3 ubiquitin ligase activity in ubiquitination assay. Subsequently, we identified PRMT4b as a PQT3-interacting protein. By histone methylation, PRMT4b upregulates the expression of APX1 and GPX1, encoding two key enzymes against oxidative stress. On the other hand, PRMT4b is recognized by PQT3 for targeted degradation via 26S proteasome. Therefore, we have identified PQT3 as an E3 ligase that acts as a negative regulator of activated response to oxidative stress and found that histone modification by PRMT4b at APX1 and GPX1 loci plays an important role in oxidative stress tolerance.


Introduction
Sessile plants cannot avoid harsh living conditions such as drought, salinity, cold and hot temperature. These stresses alter the normal cell homeostasis and increase the generation of reactive oxygen species (ROS) [1]. ROS could also be generated by paraquat, one widely used herbicide [2]. Accumulation of ROS directly destroys biological membranes and macromolecules, accelerates cell senescence, induces irreversible damages to cells and even leads to cell death [3]. The level of ROS is increased sharply under stress conditions [1,[4][5][6][7][8][9]. Two protection systems, enzymatic and non-enzymatic, have evolved to scavenge ROS and protect plant cells from oxidative damage. Enzymatic system mainly includes the ascorbate peroxidase (APX), glutathione peroxidase (GPX), catalase (CAT), superoxide dismutase (SOD), and peroxiredoxin Q (PRXQ) [8,10,11]. The transcript level and activity of antioxidant enzymes correlate with paraquat tolerance [12].
The activation of oxidative response involves many layers of regulations [8,13]. Little is known about the regulation by histone methylation of the genes involved in oxidative stress response. Histone methylation plays important roles in the plant development and growth as well as in some stress responses [14][15][16][17][18]. The methylation marks are written on lysines or arginines respectively by protein arginine methyltransferases (PRMTs) and histone lysine methyltransferases (HKMTs). In Arabidopsis, nine PRMTs are found in the genome [14]. It has been reported that PRMTs are involved in salt stress responses, flowering time as well as circadian cycle [19,20]. Two different types of PRMTs catalyze asymmetric di-methylation (ADMA) and symmetric di-methylation (SDMA) on the Arg residues, respectively [17,21]. The prmt5 mutant has decreased level of histone H4 Arg3-SDMA, leading to enhanced drought tolerance [22]. A pair of PROTEIN ARGININE METHYLTRANSFERASE4 (PRMT4) homologs, AtPRMT4a and AtPRMT4b, is required for the asymmetrical di-methylation of Arg-2, Arg-17, and Arg-26 in histone H3 [14]. The prmt4aprmt4b double mutants are sensitive to salt stress [20]. Protein arginine methylation plays essential roles in diverse biological processes, such as RNA processing and transcriptional regulation [21,23].
Oxidative stress could be perceived by multiple mechanisms, including sensor or cellular receptor. The perception by receptors results in the activation of Ca 2+ -calmodulin and mitogen-activated protein kinase (MAPK) cascade signaling transduction pathway. The activation or suppression of different transcription factors regulates a variety of defense pathway subsequently, such as ROS-scavenging, heat-shock proteins (HSPs), and photosynthesis [8,10,13]. Several paraquat tolerance mutants have been analyzed in Arabidopsis [24]. While much attention has been paid to how plants respond to oxidative stress, we know little about how plants switch off the activated responses when stress is diminished. A common regulatory mechanism is to control the protein level of the stress responsive factors. The most studied mechanism of protein degradation is the ubiquitin/26S proteasome system [25]. The ubiquitin/26S proteasome pathway is involved in different regulatory processes of eukaryotic cells, as it can rapidly eliminate the specific proteins in the cell [25][26][27]. Ubiquitin, containing 76 amino acids, could be attached to the target protein under the action of three different enzymes [27]. The ubiquitin system can identify and modify many intracellular proteins, such as proteins involved in signal transduction, transcription factors, and receptors on cell surface, to participate in the regulation of physiological processes [28]. The target proteins with ubiquitin have different fates. For monoubiquitination, one lysine residue of substrate is modified by a single ubiquitin. If several individual lysine residues of target protein are attached with single ubiquitin respectively, the protein modification is named multiubiquitination. Both mono-and multi-ubiquitination could affect protein activity and intracellular localization [29][30][31]. For polyubiquitination, one ubiquitin is attached to lysine residue of substrate firstly. The C-terminal glycine of next ubiquitin is linked with lysine residue of the preceding ubiquitin to form polyubiquitin chain subsequently [32]. As ubiquitin contains seven lysines, polyubiquitin chains are divided into different types according to different linkages between two adjacent ubiquitins [33]. Proteins with Lys48-Linkage polyubiquitination could be recognized and degraded by the 26S proteasome [34]. It has also been reported that Lys29-Linkage polyubiquitination involved in proteasome-dependent degradation [35]. In addition, Lys63-Linkage polyubiquitination plays roles in endocytosis, repair of DNA damage, protein synthesis and signal transduction [36,37].
In the ubiquitin degradation process, E3 played a crucial role. E3 is responsible for specific recognition of substrate protein and accurate positioning of the binding site between substrate protein and ubiquitin [38]. In Arabidopsis thaliana, the genes encoded the subunits of E3 ubiquitin ligases make up approximately 90 percent of about 1400 genes encoded the components of ubiquitin/26S proteasome pathway. [26,39]. The large and diverse family of plant E3 ubiquitin ligases can be divided into HECT domain-and RING/U-box domain-containing E3 ubiquitin ligases [25]. The HECT family is relatively small compared with the RING domaincontaining family that contains several hundreds of proteins and can be further divided into single subunit RING/U-box E3 ligases and multi subunit RING E3 ligases [26,40]. The large number of E3 ubiquitin ligases in higher plants indicates their important regulatory roles in diverse biological processes [41].
Several RING E3 ligases play positive roles in ABA-mediated drought tolerance [42][43][44][45]. In addition, overexpression of RING E3 ligases Rma1 inhibited the trafficking of aquaporin PIP2;1 and promote protein degradation of PIP2;1 to enhance drought tolerance [46]. A few E3 ligases have also been identified to turn off the activated stress responses [47]. HOS1, as a RING E3 ligase, acts as a negative regulator in cold tolerance. HOS1 inhibits the expression of CBFs and downstream cold-responsive genes through the degradation of ICE1 in ubiquitinproteasome pathway [48][49][50]. Both PUB22 and PUB23, as E3 ligases contained U-box, play roles as negative regulators in drought tolerance. They could recognize and degrade RPN12a, a subunit of 19S regulatory particle belonged to 26S proteasome, to affect Arabidopsis drought tolerance [51]. The negative function of PUB18 and PUB19 in ABA-mediated drought tolerance has also been reported [52]. The drought-induced AtERF53 could be degraded by RING E3 Ligase RGLG1 and RGLG2 that acts as negative regulators in drought tolerance [53]. Overexpression of salt-induced RING E3 ligase OsSIRP1 reduced salt tolerance in Arabidopsis [54]. In addition, several substrate receptors of CUL4 E3 ligases, DWAs, regulate ABA responses negatively [55,56]. The function of an enormous number of E3 ligases still remains to be identified. Here we report a novel negative regulator of oxidative stress response by PARAQUAT TOLERANCE3 (PQT3) in Arabidopsis. We identified a paraquat tolerant mutant, paraquat tolerance3 (pqt3), and cloned the gene PQT3 that encodes an E3 ligase containing RING/Ubox domain. The function of PQT3 was revealed, although PQT3 (At4g17410) had been identified in previous microarray data on salt stress response and cold acclimation [57,58]. The expression of APX1 and GPX1 was up-regulated in pqt3, while PQT3 was down regulated by oxidative stress. PQT3 was able to interact with PRMT4b. PRMT4b may catalyze histone methylation on APX1 and GPX1 chromatin and up-regulate their expressions, therefore protect plants from oxidative stress. When oxidative stress is diminished, PQT3 level increases and acts as E3 ubiquitin ligase to specifically target PRMT4b for degradation. Based on our results, PQT3 is a negative regulator that turns off the activated response of oxidative stress.

Results
Loss of At4g17410 confers the paraquat tolerance of pqt3 The mutant pqt3 was obtained after screening an activation-tagging library [59]. This library consisting of approximately 55, 000 independent lines was screened for mutants with enhanced tolerance to different stresses [60,61]. To isolate tolerant mutant to oxidative stress, we germinated seeds on MS medium with 2 μM paraquat. Growing green seedlings were selected as putative mutants and were named paraquat tolerance (pqt) because of their enhanced tolerance to paraquat. The pqt3, one of such mutants, was further characterized and marked as pqt3-1. The enhanced oxidative tolerance of pqt3-1 mutant was confirmed by germinating seeds on MS medium containing 0 or 2 μM paraquat. Based on the observation of green cotyledons, survival ratio were counted. In presence of paraquat, more than 60% pqt3-1 seeds germinated with green cotyledons but only 2% wild type seeds did, while all seeds of both wild type and pqt3-1 survived on MS medium without paraquat (Fig 1A and 1B). Genetic analysis showed that the mutation was recessive. All F1 backcross offsprings (pqt3-1 x wild type) were paraquat sensitive and F2 selfing population showed typical 3:1 segregation ratio (sensitive: resistant; 85:27, χ 2 = 0.0476). The result suggested that pqt3-1 mutant may have a more efficient mechanism of ROS scavenge, which was caused by loss-of-function mutation in a single nuclear gene PQT3 (At4g17410).
In pqt3-1 mutant, a single T-DNA insertion was located in the fourth intron of At4g17410 (S1A Fig). The exact integration site of the T-DNA right border was 803bp downstream of the ATG initiation codon of At4g17410. As a result, the expression of At4g17410 was completely disrupted as confirmed by RT-PCR analysis (S1B and S1D Fig). The expressions of its neighboring genes, At4g17390 and At4g17420, were not affected (S1B Fig). The At4g17410 locus includes 13 exons and 12 introns. By prediction, the open reading frame encodes a 91 kD polypeptide composed of 827 amino acids. Based on the conserved RING/U-box domain, this protein is predicted as an E3 ubiquitin ligase.
To further determine whether the loss of At4g17410 resulted in the enhanced oxidative tolerance of pqt3-1 mutant, we used another allele of pqt3, the T-DNA insertion mutant Salk_065409, which was ordered from Arabidopsis Biological Resource Center (ABRC) and its T-DNA insertion was confirmed by RT-PCR (S1A, S1C and S1D Fig). As the first identified pqt3 mutant was named as pqt3-1, the Salk_065409 was marked as pqt3-2. The pqt3-2 mutant showed similar enhanced oxidative tolerance to paraquat and had high survival ratio under different concentrations of paraquat treatment as pqt3-1 did (Fig 1C). The survival ratio of pqt3-1 and pqt3-2 were 50% and 20%, respectively, under 2 μM paraquat treatment, while none of the wild type seedlings survived under the same condition. In addition, both pqt3-1 and pqt3-2 showed a late-flowering phenotype (Fig 1D).
To confirm further, we generated functional complementation (FC) lines and 35Spro:PQT3 overexpression lines (S1E and S1F Fig). FC lines and 35Spro:PQT3 lines showed similar if not higher paraquat sensitivity to wild type under 2 μM paraquat treatment while the pqt3-1 and the pqt3-2 mutants displayed enhanced paraquat tolerance (Fig 1E and 1F). These results indicate that PQT3 is a negative regulator of oxidative stress tolerance and is responsible for the phenotype of pqt3 mutants.
In addition, we assayed H 2 O 2 level in the leaf with DAB staining, in which the chemical reaction between hydrogen peroxide and DAB lead to the formation of brown precipitate that indicates hydrogen peroxide distribution and oxidative damage. After 6 μM paraquat treatment for 12 or 24 hours, the result of 3,3 0 -diaminobenzidine (DAB) staining showed that the brown precipitate in the leaves of the wild type was more than that of pqt3 mutants ( Fig  1G-1R). As several stresses could cause oxidative damage to plants, the sensitivity of pqt3 mutants to other environmental stresses was analyzed subsequently. The result indicated that pqt3 mutants have enhanced tolerance to CdCl 2 , mannitol, NaCl, and drought stress (S2 Fig).

Spatiotemporal expression pattern and protein localization of PQT3
To investigate the spatiotemporal pattern of PQT3 expression, we generated PQT3pro:GUS reporter lines. GUS staining results showed that PQT3 was expressed in both shoot and root tissues under normal condition (Fig 2A-2G). GUS expression was detected in the root tissues at all developmental stages we analyzed (Fig 2A-2C). For the 1-week-old seedlings, strong GUS staining was observed in the cotyledons, hypocotyls and root tissues (Fig 2A). For the 3-week-old seedlings, strong GUS staining was also detected in cotyledons, young leaves, and root tissues, but weakly stained in older leaves ( Fig 2B). In 7-week-old adult plants, GUS expression was detected in rosette leaves, cauline leaves, the tip and basal junction of siliques, and was significantly higher in the flower petals, stamens and stigma of pistil (Fig 2D-2G).
PQT3 has two predicted nuclear localization signals (NLSs) in the carboxyl terminus (S3 Fig), implicating its nuclear localization. To confirm this, the 35Spro:PQT3-GFP construct was made and transiently expressed in onion epidermal cells. The 35Spro:GFP construct was used as control (Fig 2H-2J). PQT3-GFP signal was indeed detected in the nucleus (Fig 2K-2M).
For further confirmation, we transformed the PQT3pro: PQT3-GFP fusion construct into the Arabidopsis and obtained transgenic plants. Fluorescent microscopy results showed that GFP signal was accumulated in the nucleus of root cells (Fig 2N-2P), which is in agreement with the presence of two NLSs of the PQT3.

PQT3 is rapidly down-regulated by stress treatments
The expression of PQT3 was rapidly down-regulated by paraquat treatment and maintained at a low level as long as the paraquat treatment was applied ( Fig 3A). This result is supportive for our previous opinion that PQT3 is a negative regulator of plant oxidative tolerance. The suppressed transcript level of PQT3 was restored to the previous level when PQ stress diminished (S4 Fig). By extrapolation, the expression of PQT3 could be down-regulated by other stress conditions. Indeed, our results showed that the expression of PQT3 was down-regulated by H 2 O 2 (Fig 3B), mannitol (Fig 3C), drought ( Fig 3D) and CdCl 2 (Fig 3E) at the indicated time points. Among these stresses, CdCl 2 treatment led to the most significant reduction of the expression of PQT3 (Fig 3E). The expression of PQT3 could also be down-regulated by NaCl stress at 3h. However, unlike the above results, the expression of PQT3 was activated by salt treatment at other time points (Fig 3F).  under normal conditions compared with that in the wild type (Fig 4A-4J). The elevated transcript levels of APX1 and GPX1 may contribute to the improved oxidative tolerance.
The enzyme activity of APX and GPX in wild type and pqt3 mutant was also detected. The pqt3 mutant had higher enzyme activity of APX and GPX compared with wild type (Fig 4K  and 4L).

PQT3 interacts with PRMT4b
To study the molecular mechanisms that underlie the enhanced stress tolerance of pqt3, we screened cDNA library for potential candidate target proteins of PQT3 using yeast-two-hybrid (Y2H). Several proteins were isolated from the screen. Among these candidate interactors, PRMT4b, a member of arginine methyl transferase family, was frequently presented. To reveal the domain of PQT3 responsible for the interaction with PRMT4b, the PQT3 protein was divided into four parts: N-terminal DWNN, zfCCHC, U-box (RING finger), and C-terminal section containing the NLS1 and NLS2 domains, based on the predicted domains of PQT3 protein (S3A Fig). Full-length PQT3 and four protein sections were used for Y2H assay as baits. Full-length protein and the C-terminus of PQT3 were able to interact with PRMT4b in Y2H assays ( Fig 5A). The interaction between PQT3 and PRMT4b was confirmed by colonies that grew on the SD-Leu-Trp-His plate with 50 mM 3-amino-1, 2, 4-triazole (3-AT) and displayed the blue color in X-gal assay ( Fig 5A). However, PRMT4a was never isolated in the screening. The Y2H assay was also performed to further study the potential interaction between PQT3 and PRMT4a, since PRMT4a is a close related gene of PRMT4b. The result showed that the PQT3 did not interact with PRMT4a (S5 Fig).
The interaction between PQT3 and PRMT4b was further confirmed by protein pull-down assay in vitro. MBP-PQT3-C66 protein containing the NLS domain (S3A Fig) and His-PRMT4b protein were expressed in E. coli and purified subsequently. His-PRMT4b was incubated with amylose resin bound with recombinant MBP-PQT3-C66 protein. Pulled-down protein complex was detected by SDS-PAGE ( Fig 5B) and western blotting using anti-His antibody ( Fig 5C). The pull-down result clearly shows that PQT3 interacts with PRMT4b in vitro.
To determine whether the interaction also occur in vivo, we used the bimolecular fluorescence complementation (BiFC) system. The N-terminus of yellow fluorescent protein (YFP) was fused to full-length PQT3 cDNA, while full-length PRMT4b cDNA was fused to the C-terminal region of YFP. The empty plasmids were used as negative controls. Different plasmid combinations were co-infiltrated into epidermal cell of N. benthamiana leaves. The yellow fluorescence was observed in epidermal cell contained both NE-PQT3 (the N-terminus of YFP fused with PQT3) and CE-PRMT4b (the C-terminus of YFP fused with PRMT4b) (Fig 5D). No fluorescence was observed from the negative controls (NE-PQT3/CE, NE/CE-PRMT4b and NE/CE) ( Fig 5D). The nuclei were stained by Hoechst and detected by confocal. These results indicate that PQT3 can interact with PRMT4b in the nucleus of plant cell.

PQT3 has E3 ubiquitin ligase activity and can ubiquitinate PRMT4b
Not all the proteins with the predicted RING domain function as an ubiquitin ligase [62]. The E3 activity of PQT3 was determined via self-ubiquitination system. Both full-length (GST-PQT3) and C-terminal deletion (GST-PQT3-N40) proteins showed the E3 ubiquitin ligase activity ( Fig 6A and S3B Fig). The ubiquitinated bands of PQT3 were detected by western blotting in the presence of E1 (from wheat), E2 (UBCh5b, from human), and 6×His-tagged ubiquitin (UBQ14, from Arabidopsis). When any of essential reaction components was missing, self-ubiquitination of PQT3 was not detected (Fig 6A).
To determine whether PRMT4b is a substrate recognized by PQT3, we resorted to the in planta ubiquitination assay [63]. Leaf infiltration was conducted via Agrobacterium tumefaciens strains containing different combination of constructs. The infiltrated parts of N. benthamiana leaves were harvested. Total protein was extracted and detected via western blotting with anti-HA antibody. A smear of bands, which were larger than the size of HA-PRMT4b and showed the features of ubiquitinated form of the PRMT4b proteins, could be detected by anti-HA antibody in the samples co-infiltrated with PQT3 and HA-PRMT4b ( Fig 6B). The cell lysates were immunoprecipitated with anti-HA antibody subsequently. Immunoprecipitated samples were detected via western blotting with anti-ubiquitin antibody. In the PQT3-PRMT4b co-infiltration sample, these high molecular size bands could also be detected by anti-ubiquitin antibody ( Fig 6B). These results indicated that these high molecular size bands were ubiquitinated forms of PRMT4b. PQT3 protein could ubiquitinate the PRMT4b protein in tobacco. The decline of PRMT4b protein was also found in the samples co-infiltrated with PQT3 and HA-PRMT4b (Fig 6B).
In addition, PRMT4b protein levels in different PQT3 genetic background under MG132 treatments were also supportive of PRMT4b as a substrate of PQT3. As shown in the Fig 6C  and 6D, the protein level of PRMT4b in pqt3-1 was higher than that of the wild type under normal conditions. Under paraquat treatment, it remained lower in wild type than that of pqt3-1 at the same time point, although protein level of PRMT4b increased gradually in wild type and pqt3-1 with the prolonged paraquat treatment (Fig 6C and 6D). The transcript level of PRMT4b was unlikely to cause the observed difference of protein levels between the mutant and wild type (S6 Fig). Under MG132 treatment, the protein level of PRMT4b was increased in both wild type and pqt3-1 mutant. When the seedlings were co-treated with paraquat and MG132 for 12 h, PRMT4b was accumulated in both wild type and pqt3-1 mutant, and no significant difference of PRMT4b protein level was found between the wild type and pqt3-1 mutant (Fig 6C and 6D) because the degradation of the PRMT4b protein through ubiquitination-26S proteasome pathway was inhibited by MG132.
In order to confirm further the PQT3-dependent ubiquitination of PRMT4b, the phenotype of wild type and pqt3-1 mutant treated without or with paraquat in presence or absence of proteasome inhibitor MG132 was studied. Survival ratio of pqt3-1 seedlings was higher than that of wild type under paraquat treatment. No significant difference of survival ratio could be observed when the wild type and pqt3-1 seedlings were co-treated with paraquat and MG132 (Fig 6E). The wild type gained enhanced paraquat tolerance under MG132 treatment to the level of pqt3-1 mutant.
PRMT4b is a positive regulator in the oxidative stress tolerance To reveal whether PRMT4b could play any roles in oxidative tolerance of plants, we obtained the prmt4b mutant and 35Spro:PRMT4b lines (S1G, S1H and S1L Fig) and observed the phenotypes of prmt4b and 35Spro:PRMT4b under different concentrations of paraquat treatment. Survival ratio of wild type, pqt3-1, pqt3-2, prmt4b and 35Spro:PRMT4b were counted ( Fig 7A). 35Spro:PRMT4b had similar phenotype as pqt3-1 and pqt3-2, while the prmt4b mutant was more sensitive to paraquat treatment than wild type ( Fig 7A). Under CdCl 2 treatment, primary root elongation of prmt4b mutant was also slower than that of wild type ( Fig 7B). Furthermore, the overexpression lines of PRMT4b were analyzed under other stress conditions. 35Spro:PRMT4b increased tolerance to CdCl 2 and NaCl stresses compared with wild type. The 35Spro:PRMT4b showed the opposite phenotype of the 35Spro:PQT3 which was more sensitive to CdCl 2 and NaCl stresses as compared with wild type (S7 Fig). These results show that PRMT4b is a positive regulator for plant oxidative tolerance, which is also consistent with the function of PQT3. The prmt4a mutant and prmt4aprmt4b double mutants were subsequently examined (S1I-S1K Fig). The prmt4aprmt4b double mutants had the similar phenotype to prmt4b under paraquat treatment, and the knockout of PRMT4a did not affect the oxidative tolerance (Fig 7C-7E). In addition, the transcript levels of APX1, GPX1 and other antioxidant enzyme genes were down-regulated in prmt4b mutant and upregulated in 35Spro:PRMT4b as compared with that in the wild type ( Fig 7F and S8 Fig).
Increased H3R17 methylation on APX1 and GPX1 chromatin in pqt3 The transcript level of APX1 and GPX1 was higher in pqt3 than that in wild type under normal conditions (Fig 4A and 4B), which could be regulated by PRMT4b. The modification status of H3R17me2a in the chromatin of APX1 and GPX1 was compared between wild type and pqt3 mutant via chromatin immunoprecipitation (ChIP) assays. ChIP assays were performed with wild type and pqt3 plants using antibody against H3R17me2a. As shown in the Fig 8A and 8B, the APX1 and GPX1 chromatin was divided into different regions and the enriched chromosome fragments were detected by quantitative RT-PCR. The results showed that histone H3R17me2a modification of APX1 and GPX1 chromatin was increased in pqt3 mutant (Fig 8C  and 8D). The enriched chromosome fragments of APX1 (C, D and E fragments) and GPX1 (C, D and I fragments) were further analyzed in prmt4b mutant and pqt3prmt4b double mutants without or with paraquat treatment (Fig 8E-8J). Histone H3R17me2a modification of APX1 (C, D and E fragments) and GPX1 (C, D and I fragments) chromatin was decreased in prmt4b mutant and pqt3prmt4b double mutants without paraquat treatment as compared with wild type. It remained lower in wild type than that of pqt3 mutant, although H3R17me2a modification of APX1 (C, D and E fragments) and GPX1 chromatin (C, D and I fragments) in both wild type and pqt3 mutant were enhanced under paraquat treatment. H3R17me2a modification of APX1 (C, D and E fragments) and GPX1 chromatin (C, D and I fragments) was kept at a low level, although the modification in prmt4b mutant and pqt3prmt4b double mutants could also be affected by paraquat treatment. These results suggest that PRMT4b may target APX1 and GPX1 to enhance the oxidative tolerance by increasing asymmetric dimethylation of H3 at R-17 in APX1 and GPX1 chromatin.

pqt3prmt4b double mutants in oxidative tolerance
To confirm further that PRMT4b is the target of PQT3, the pqt3prmt4b double mutants were obtained (S1M and S1N Fig). The survival ratio of pqt3prmt4b under paraquat treatment was intermediate between pqt3 and prmt4b, which demonstrates that the PRMT4b protein is one of the targets of PQT3 and suggests that PQT3 may also target other proteins that contribute to the tolerance to oxidative stress (Fig 9A-9C).
We also obtained the pqt3prmt4a double mutants (S1O and S1P Fig) and found that the survival ratio of pqt3prmt4a under paraquat treatment had no significant difference from that of pqt3 mutant (Fig 9D-9F), again indicating that PRMT4a is not involved in oxidative stress response.

Discussion
PQT3 is a member of Arabidopsis RING-finger/U-box E3 ligase family. Secondary structure prediction using InterProScan protein sequence analysis software revealed four conserved Fig 6. PRMT4b is a substrate recognized by PQT3. (A) In vitro ubiquitination assay. GST-PQT3-N40 and GST-PQT3 were expressed in E. coli and purified. Nickel-HRP was used to detect His-tagged ubiquitin. The ubiquitination activity of PQT3 was observed only in the presence of E1 (from wheat), E2 (UBCh5b, from human), and 6×His-tagged ubiquitin (Ub). GST was used as negative control. The numbers on the left indicate the molecular masses of marker proteins. (B) In planta ubiquitination assay. Total protein samples were isolated from the infiltrated parts of N. benthamiana leaves 1 day after agroinfiltration. The total protein was analyzed via western blot using anti-HA antibody (upper panel). Immunoprecipitated samples were analyzed using western blot with anti-ubiquitin antibody subsequently (middle panel). Ponceau S staining of the Rubisco protein was shown as a loading control (lower panel). The numbers on the left indicate the molecular masses of marker proteins. (C and D) The effect of proteasome inhibitor MG132 on the protein level of PRMT4b in wild type and pqt3-1 mutant without or with paraquat treatment in time-course. Wild type and pqt3-1 mutant were grown for 14 days, then the seedlings were treated without or with 6 μM paraquat for 6 or 12 h in presence or absence of 50 μM proteasome inhibitor MG132. The total proteins were extracted. Western blot was performed using anti-PRMT4b antibody (C). Ponceau S staining of the rubisco protein serves as a loading control. Band densities were quantified using Quantity One software (Bio-Rad, USA) (D). Values are mean ± SD (n = 3 experiments, ***P < 0.001). Asterisks indicate Student's t-test significant differences. (E) The survival ratio of wild type and pqt3 mutant germinated and grown on MS medium containing 0 μM or 1 μM paraquat without or with 15 μM MG132 for 7 days were counted. DMSO was used as control. Values are mean ± SD (n = 30 plants, *P < 0.05). Asterisk indicates Student's t-test significant difference.  [64]. The DWNN domain of PQT3 contains 76 amine acids. DWNN domain is only found in the N-terminus of the members in the splicing-associated RBBP6 (Retinoblastoma Binding Protein 6) protein family [64]. The RING-finger domain was also found in RBBP6 protein family. The existence of the RING-finger domain suggests that DWNN domain may play its role as ubiquitin-like regulatory factors [64]. CCHC-type Zinc finger domain is also known as zinc knuckle, which can be found in a large number of RNA binding proteins [65,66]. As mentioned above, RING-finger domain can combine with the E2 in the cascade reaction of ubiquitination system,while U-box is a modified RING-finger [67]. In addition, two predicted NLS sequences (471-477 and 696-711 amino acids) were also found in the C-terminus of PQT3, which is consistent with nuclear localization of the protein (Fig 2H-2P and S9A and S9B Fig). In vitro ubiquitination assay shows that PQT3 has the E3 ligase activity (Fig 6A). We noticed that the in vitro activity was not as high as expected, which could be contributed by suboptimal reaction conditions such as E2 source. This E3 activity was confirmed by in planta ubiquitination assay, in which PRMT4b protein could be ubiquitinated by expressed PQT3 protein in N. benthamiana (Fig 6B). Consistent with this result, pqt3-1 mutant had higher protein level of PRMT4b than the wild type, as the PRMT4b protein was degraded by PQT3 in wild type under normal conditions (Fig 6C and 6D). The transcript level of PRMT4b could be induced in wild type under paraquat treatment (S6 Fig), while the transcript level of PQT3 was decreased by paraquat treatment (Fig 3A). In this case, PQT3-mediated PRMT4b degradation could be weakened in wild type. Consequently, the level of PRMT4b protein in wild type was elevated, but lower than that of pqt3 (Fig 6C and 6D). The proteasome inhibitor MG132 could block the degradation of the PRMT4b protein in wild type and enhance paraquat tolerance of wild type (Fig 6C-6E). Taken together, PQT3, as an E3 ubiquitin ligase, play its role in oxidative tolerance through the ubiquitination-degradation of PRMT4b.
A series of environmental stress could lead to oxidative damage to plants [10,13,68]. Both biotic stress and abiotic stress result in the production of ROS which in excess cause oxidative stress. Besides paraquat tolerance, pqt3 mutants have enhanced tolerance to various environment stresses (Fig 1 and S2 Fig). Further studies may reveal new mechanisms of PQT3 in multiple stress tolerance. Analysis of PQT3-interacting proteins may be a good start point to further understand the function of PQT3. In the screen for PQT3-interacting proteins with Y2H, we isolated 66 positive colonies based on the expression of reporter genes included His3

Fig 8. Arg-17 methylation in specific regions of APX1 and GPX1 chromatin is increased in pqt3 mutant. (A and B) The illustration of APX1 and GPX1 chromatin. A to F represent different regions of APX1 chromatin (A). A to I represent different regions of GPX1 chromatin (B). Promoters (black lines), exons (white boxes), and introns (black lines) of APX1
and GPX1 were shown. (C and D) ChIP-PCR assay. Quantitative PCR was performed to verify each chromatin region of APX1 and GPX1 using specific primers. Fragments C, D and E in APX1 chromatin were enriched by anti-H3R17 antibodies in pqt3 mutant (C). Fragments C, D and I in GPX1 chromatin were enriched by anti-H3R17 antibodies in pqt3 mutant (D). UBQ5 was used as an internal control. Values are mean ± SD (n = 3 experiments, **P < 0.01, ***P < 0.001). Asterisks indicate Student's t-test significant differences.(E-G) The enrichment of C fragment (E), D fragment (F) and E fragment (G) in APX1 chromatin was analyzed in wild type, pqt3 mutant, prmt4b mutant and pqt3prmt4b double mutants without or with paraquat treatment. UBQ5 was used as an internal control. Values are mean ± SD (n = 3 experiments, **P < 0.01, ***P < 0.001). Asterisks indicate Student's t-test significant differences. (H-J) The enrichment of C fragment (H), D fragment (I) and I fragment (J) in GPX1 chromatin was analyzed in wild type, pqt3 mutant, prmt4b mutant and pqt3prmt4b double mutants without or with paraquat treatment. UBQ5 was used as an internal control. Values are mean ± SD (n = 3 experiments, **P < 0.01, ***P < 0.001). Asterisks indicate Student's t-test significant differences. doi:10.1371/journal.pgen.1006332.g008 and LacZ. Several proteins were identified from these positive colonies repeatedly. The most frequent interacting partner is 20S core protease subunit of 26S proteasome. The result suggested that PQT3 act as an E3 ubiquitin ligase. Among these proteins, PRMT4b, as previously mentioned, may be responsible for the increased degree of Arg-17 methylation which may further regulate APX1 and GPX1 genes to enhance oxidative tolerance of plants (Figs 4A, 4B and  8). It has been reported that arginine methylation was involved in signal transduction, transcriptional control, DNA repair, RNA processing, and nuclear transport [17,19,23,[69][70][71][72]. The functions of PRMTs have been extensively analyzed [15,19,20]. Here, the prmt4b mutant was found to be more sensitive to paraquat and CdCl 2 treatment than wild type (Fig 7A-7E). Thus, PRMT4b plays a role in oxidative stress response of plants. It is known that PRMT4-mediated methylation at Arg-17 of histone H3 is linked to transcription activation [73]. The Arg-17 of histone H3 is the major site of PRMT4-mediated methylation, although it was reported that other protein site could also be methylated by PRMT4 [74][75][76]. We suggest that the increased Arg methylation degree of specific regions in APX1 and GPX1 chromatin was caused by PRMT4b (Fig 8).
The increased transcript level of APX1 and GPX1 may be caused by PRMT4b-mediated histone Arg methylation (Figs 4A, 4B and 8). In presence of ascorbate as electron donor, the cytosolic enzyme, APX1, catalyze the degradation of H 2 O 2 [77]. The response of APX1 to oxidative stress has been studies in Arabidopsis. The crucial role of APX1 in multiple stress response was reported [78]. It could be activated by multiple stresses to protect plants against oxidative stress [79][80][81]. The tobacco overexpressing APX1 could be more tolerant against UV-C-caused oxidative damage [82]. GPXs also have important functions in oxidative signaling, which can protect plants from harmful effects of excessive oxidation [83]. It has also been reported that enhanced peroxide scavenging and decreased oxidative damage was found in transgenic tobacco seedlings overexpressing tobacco glutathione S-transferase that showed glutathione peroxidase (GPX) activity [84].
The pqt3 mutants also have late-flowering phenotype (Fig 1D). It has been reported that prmt4aprmt4b double mutants display late-flowering phenotype [15]. The Y2H result showed that the PQT3 can not interact with PRMT4a ( S5 Fig). As compared with wild type, the prmt4a mutant also has no significant difference in the oxidative tolerance (Fig 7C-7E). The phenotype of pqt3prmt4a double mutants demonstrate that PRMT4a protein was not involved in the response regulation of oxidative stress by PQT3 (Fig 9D-9F). The late-flowering phenotype of pqt3 may be regulated by other mechanisms, rather than PRMT4a and PRMT4b. The flowering-related transcription factor AGAMOUS (AG) was found to be a potential target interacted with PQT3 in Y2H library screening. AG is involved in carpel development, leaf development, identification of floral organs, and stamen development [85]. Targeted removal of AG by PQT3 may be related to the late-flowering phenotype of pqt3. In addition, one member of a large G protein family and the proteins with unknown function were also found to be potential targets interacted with PQT3 in Y2H library screening. By interacting with different partners, PQT3 may mediate multiple functions in diverse biological processes.
The pqt3 has enhanced tolerance to multiple stresses. PQT3 is also down-regulated by various oxidative stresses. PQT3 act as negative regulator in multiple stress responses. The interacting partner will be further analyzed to reveal the molecular function of PQT3. As research continues, other combinations, also targeted by PQT3, may be revealed, where PQT3 act as a positive regulator. We could improve crop tolerance to multiple stresses through PQT3 mutation. More candidate genes could be further determined from the potential targets interacted with PQT3. These genes may also have high application value after the revelation of molecular mechanisms.
In conclusion, paraquat may enter the cell through plasma membrane-localized paraquat transporters PDR11 and RMV1 [59,86]. The intracellular paraquat may further activate different downstream signaling pathways to regulate the expression of PQT3 and PRMT4b. In addition, paraquat stress increases the generation of ROS. ROS is also an important signal molecule that mediate the responses to environmental stress [87]. ROS-activated signaling pathways may also be responsible for the regulation of PQT3 and PRMT4b. The oxidative stress activates the expression of PRMT4b, represses the expression of PQT3, and weakens PQT3-mediated the ubiquitinated degradation of PRMT4b, synergistically resulting in increased accumulation of PRMT4b. Consequently, the increased level of PRMT4b protein may lead to higher degree of histone methylation on the APX1 and GPX1 chromatin. As a result, the transcription of APX1 and GPX1 is activated, leading to more APX1 and GPX1 enhancing oxidative tolerance of plants. When the stress disappears, transcription repression of PQT3 by oxidative stress is removed. The function of PQT3, as a negative regulator of oxidative stress response, is restored. The PRMT4b is then degraded by PQT3 in ubiquitination pathway. The activated response to oxidative stress is switched off. We propose a working model for PQT3 as a negative regulator of oxidative stress response (Fig 10).

Mutant screen from the activation-tagging library
Approximately 55,000 individual lines were screened for paraquat tolerant mutant, which came from the activation-tagging library constructed with Arabidopsis Columbia ecotype and pSKI015 vector [59].

Plant material and growth conditions
All Arabidopsis transgenic lines and mutants were based on the genetic background of Columbia ecotype used as wild type in the study. Salk_065409 (pqt3-2), Salk_097442C (prmt4b) and Salk_033423 (prmt4a) were obtained from ABRC (Arabidopsis Biological Resource Center). The seeds were sterilized in 10% bleach for 10 min. Then the seeds were washed for 5 times at least with sterile water. For vernalization, the seeds were kept in the dark with water at 4°Cfor 3 days to ensure the synchronous germination. Sterile seeds were germinated on MS medium. The seedlings were grown at 22°C under 16-h-light /8-h-dark cycle with light intensity 100 μE m -2 s -1 .

Transformation of Arabidopsis
The constructs were electroporated into competent cell of Agrobacterium tumefaciens C58C1. The floral-dip method was used to transfer these constructs into Arabidopsis as described [88,89].

PCR analysis
The different tissues of plants were used for RNA extraction with Trizol method. Then RNA reverse reaction was carried out by TransScript Kit (TransGen Biotech). Specific primers were designed for RT-PCR analysis and the PCR products were detected by agarose gel electrophoresis. Applied Biosystem Step One real-time PCR system was used for the quantitative RT-PCR detection with specific primers listed in the S1 Table and Premix Ex Taq II SYBR (TaKaRa). UBQ5 was used as the internal control.
The identification of homozygous mutants was performed by genomic PCR for T-DNA insertion lines: Salk_065409, Salk_097442C and Salk_033423 [90]. RT-PCR was carried out as previously described to confirm the results of genomic PCR. The 35Spro:PQT3 and 35Spro: PRMT4b plasmids were transformed into Col-0 to obtain the overexpression lines of PQT3 and PRMT4b. For the FC line, the 35Spro: PQT3 was transformed into pqt3 mutant, and the line with the same expression level of PQT3 as wild type was chosen and used as FC line. 35Spro: PQT3, 35Spro: PRMT4b and FC line were identified by glufosinate screening and quantified by RT-PCR or quantitative RT-PCR.

Stress tolerance assay
The seeds were germinated on MS medium containing different concentrations of paraquat, mannitol, CdCl 2 and NaCl, respectively. The phenotype was observed and survival ratio was scored at the indicated time points.
For drought tolerance assay, the wild type, pqt3-1 mutant and pqt3-2 mutant seeds were germinated in one pot at same density. The 15-day-old seedlings were used for drought tolerance Under oxidative stress, stress signaling up-regulates PRMT4b expression and down-regulates PQT3 expression, leading to higher PRMT4b activity that will activate APX1 and GPX1 and enhance antioxidation capacity of APX1 and GPX1. When oxidative stress is diminished, PRMT4b expression is decreased and PQT3 expression is increased. As a result, PQT3 activity is increased, leading to faster removal of PRMT4b via 26S proteasome. Together with decreased PRMT4b expression, PRMT4b activity will rapidly drop, leading to decreased expression of APX1 and GPX1. The activated response of oxidative stress is then switched off.
doi:10.1371/journal.pgen.1006332.g010 assay. Watering was withheld for another 15 days before re-watering. The photos were taken before re-watering and after re-watering for 1 day and 7 days. The survival ratio was scored after re-watering for 1 day and 7 days.

DAB staining
DAB staining was performed as described [91]. DAB staining solution (pH 6.0) was prepared by adding 0.05% (v/v) Tween-20 and 10 mM Na 2 HPO 4 to the DAB solution (1 mg/ml DAB, pH3.0). For each treatment condition, at least 3 leaves per plant were obtained from 3 independent plants for each line (Col-0, pqt3-1 and pqt3-2). Arabidopsis leaves from different lines were treated using MS liquid medium without or with 6 μM paraquat for 12 hr or 24 hr. These leaves were stained in 6-well culture plates with DAB staining solution subsequently and 10 mM Na 2 HPO 4 (pH 6.0) was used as the negative control. The 6-well plates were covered with aluminum foil and placed on a shaker for 4-5 h. Follow the incubation, the bleaching solution (glycerol: acetic acid: ethanol = 1:1:3) was used in the discoloration after DAB staining solution was removed. The 6-well plates were placed into boiling water bath at 95°C for 15-20 min. Discoloration process was repeated using fresh bleaching solution. The brown precipitate caused by the reaction between DAB and H 2 O 2 could be observed on the leaves. Photos were taken using a camera. Special attention should be paid to light avoidance through the whole operation. The experiment was repeated for three times.

The detection of GUS activity
The promoter of PQT3 was cloned into pCB308R [92,93]. The transgenic lines containing PQT3pro: GUS were isolated by glufosinate screening. The T2 population was used for GUS staining. Histochemical staining for GUS activity in Arabidopsis was carried out as described previously [94] and GUS staining solution was prepared as described before [59]. The experimental materials were incubated in staining solution at 37°C. Then Arabidopsis tissues were destained and stored in 70% ethanol. The GUS activities of individual parts were observed via a light microscope Axio skop2 plus (ZEISS, Germany) with a video camera.

Subcellular localization assay
The full length CDS of PQT3 was cloned into the binary vector pCB2008E to construct the PQT3-GFP fusion vector [93]. After the detection of gene sequencing for inserted sequence, the recombinant plasmid was transducted into the epidermal cells of onion along with particle gun bombardment for transient expression assay. PQT3-GFP was also transferred into Col-0 via floral-dip method to create transgenic plant for the analysis of PQT3-GFP fusion protein localization. ZEISS fluorescence microscope (Axio skop2 plus) with video camera was used to observe the green fluorescence in both the epidermal cells of onion and the root cells of Arabidopsis transgenic lines.

Enzyme activity assay
For APX and GPX enzyme activity assay, Arabidopsis seedlings were ground in liquid nitrogen and resuspended in precooling Enzyme extraction buffer (50 mM phosphate buffer (Na 2 H-PO 4 -NaH 2 PO 4 ), pH = 7.0; 2% (w/v) polyvinylpolypyrrolidone (PVPP); 0.05% (v/v) TritonX-100; 1 mM EDTA and 1 mM ascorbic acid) on ice. Extraction solution was centrifuged for 20 min (16,000 g/min, 4°C). The protein concentration of supernatant was detected by Onedrop micro-ultraviolet spectrophotometer and SDS-PAGE before the supernatant was used for enzyme activity analysis. The detection of APX activity was performed as described with modifications [95]. For APX activity, 50 μl enzyme and 2950 μl reaction mixture (50 mM Tris-HCl, pH7.0; 0.1 mM EDTA; 0.1 mM H 2 O 2 and 0.5 mM ascorbic acid) was mixed. The decreased OD 290 was recorded per 10 s. The enzyme amount oxidized 1 mM AsA in one minute set as one activity unit (U) of APX. The APX activity was defined as U • g -1 protein. GPX activity was measured indirectly through the detection of glutathione reductase (GR) activity using GPX Activity Measurement Kit (Beyotime Biotech, China). GR activity was detected as described with modifications [96]. The OD 340 was recorded per 30 s. The enzyme amount consumed 1 mM NADPH in one minute set as one activity unit (U) of GPX. The GPX activity was also defined as U • g -1 protein.

In vitro E3 ubiquitin ligase activity assay
The in vitro E3 ubiquitin ligase activity assay was carried out as described previously [43]. GST-PQT3 fusion protein was obtained from E. coil and purified subsequently. His-tagged ubiquitin of Arabidopsis (UBQ14) was also expressed using bacterial expression system and purified. In addition, the wheat (Triticum aestivum) E1 (GI: 136632) and human E2 (UBCh5b) were also used in the reaction. Reactions were performed for 1.5 h at 30°C. For the immunoblot, His-tagged ubiquitin of Arabidopsis was detected using Nickel-HRP (Kirkegaard & Perry Laboratories, nickel-nitrilotriacetic acid agarose conjugated to horseradish peroxidase).

Y2H screening and confirmation
The bait plasmid was transformed into Mav203 strain of yeast, which was constructed with pDEST32 vector and the full-length PQT3 cDNA. Yeast two-hybrid screening was performed using two-hybrid cDNA library of Arabidopsis. The cDNA library containing fragments of fusion proteins composed of prey proteins and GAL4-AD was used to transform Mav203 cells harboring bait plasmid in Y2H assay. Positive clones were screened using SD/-Leu-Trp-His and X-gal assay. Then it was identified by nucleotide sequencing with corresponding primers. Twohybrid screening was carried out based on the protocol described in Two-Hybrid System Manual (Invitrogen, USA). The result of screening was further confirmed by the two-hybrid assay. The full-length CDS of PQT3 and the four segments of PQT3 were inserted into pDEST32 vector to construct the bait plasmid, while the prey plasmid was constructed with pDEST22 vector and the full-length CDS of PRMT4b. For two-hybrid assay, the primers could be found in S1 Table. Pull-down assay MBP-PQT3-C66 and His-AtPRMT4b fusion protein were expressed using prokaryotic expression system and purified. MBP-PQT3-C66 fusion protein was incubated with MBP beads (amylose resin) at 4°C for 2 h, and the MBP tag was used as a negative control. The beads were cleaned with washing buffer for 4 times. Then the beads were incubated with His-AtPRMT4b at 4°C for 2 h respectively. The beads were cleaned with washing buffer for 4 times. Western blot was used to detect the SDS-PAGE separation results of pulled-down mixtures in nitrocellulose membrane with anti-His antibody.

Agroinfiltration procedure
Agrobacterium tumefaciens strain C58C1 was used in the experiments. Agroinfiltration procedure was carried out as described previously [63]. At first, these strains were grown on LB medium with Gentamicin and Kanamycin. Single colony was transferred into 5 ml LB liquid medium containing same resistance and grown for 48

BiFC analysis
NE-PQT3 (the N-terminus of YFP fused with PQT3) and CE-PRMT4b (the C-terminus of YFP fused with PRMT4b) were constructed. These constructs were transferred to Agrobacterium strains C58C1 respectively. As mentioned above, the different plastid combinations were transformed into epidermal cells of N. benthamiana leaves by agroinfiltration. YFP was observed 1-2 days after leaf infiltration using confocal. The nuclei were stained by Hoechst subsequently and fluorescence detection by confocal was performed. For BiFC assay, the primers could be found in S1 Table. Protein extraction Native extraction buffer 1 [10 mM EDTA; 50 mM TRIS-MES, pH 8.0; 1 mM MgCl 2 ; 5 mM DTT; 0.5 M sucrose; protease inhibitor cocktail for plant cell and tissue extracts (Sigma, USA)] was chosen for protein extraction buffer. Other steps of protein extraction were carried out as described previously [63].
In planta ubiquitination assay N. benthamiana leaves was infiltrated using the mixture of Agrobacterium strains containing different constructs, from which total proteins were extracted 1 day after Agroinfiltration. The p19 was used as gene-silencing suppressor. The anti-HA antibody was used for western blot detection of the total proteins after the separation of SDS-PAGE. HA-PRMT4b is a fusion protein composed of HA-tag and PRMT4b.The protein level of PRMT4b could be analyzed via the HA-tag detection using anti-HA antibody. Ponceau S staining of the Rubisco protein was used as loading control. In the presence of MG132, total proteins were immunoprecipitated with protein A agarose beads (Millipore, USA) and anti-HA antibody (Abmart) subsequently. Detailed steps of immunoprecipitation were carried out as described previously [63]. Immunoprecipitated samples were analyzed using western blot with anti-ubiquitin antibody (Abcam). Image Quant LAS 4000 (GE, USA), as the CCD camera system, was used for the result examination.

Chromatin immunoprecipitation-PCR assay
The wild type, pqt3 mutant, prmt4b mutant and pqt3prmt4b double mutants were used for ChIP assay without or with 6 μM paraquat treatment for 24h. UBQ5 was chose for internal control. ChIP was performed as previously described [97,98]. The regions with Arg-17 methylation were precipitated via anti-H3R17me2a antibodies (anti-Histone H3 asymmetric dimethyl R17 antibody-ChIP grade, ab8284, Abcam, USA) from input DNA. The corresponding primers were designed for quantitative RT-PCR to detect the enrichments of different DNA fragments in APX1 and GPX1 chromatins [99,100]. The primers used in ChIP assay were showed in S1 Table. Supporting Information The division of PQT3 protein for Y2H assay was presented by the red lines. The full-length protein sequence of PQT3 was divided into four segments (DWNN, zfCCHC, U-box/RING finger and C-terminus contained the NLS1 and NLS2 domain) and each segment was used as bait. In pull-down assay, the protein section of PQT3 (PQT3-C66) located in the blue box was used. (B) The protein used in Self-ubiquitin assay. Full-length protein sequence of PQT3 and PQT3-N40 (the section of PQT3 protein located in the red box) were selected for the assay.