The rice SCFOsFBK1 E3 ligase mediates jasmonic acid induced degradation of a RING-H2 protein and the cinnamoyl-CoA reductase, OsCCR14


 We had previously shown the rice F-box, OsFBK1, plays a role in anther development by mediating the turnover of OsCCR14, a cinnamoyl CoA-reductase regulating lignification. Another substrate identified in the same Y2H library screening was OsATL53, a member of the ATL family of RING-H2 proteins that is primarily localized to the cytoplasm. We found OsATL53 to be a component and substrate of SCFOsFBK1 by immunoprecipitation and cell-free studies. Incidentally, OsATL53 was found to interact with OsCCR14 in the cytoplasm and form a stable complex in cell-free experiments and bimolecular fluorescence complementation assays. Biochemically, OsATL53 was found to influence the enzymatic activity of OsCCR14 by decreasing its efficiency. Degradation studies have shown OsFBK1 mediates turnover of OsCCR14 in the nucleus, while OsATL53 is degraded in both cytoplasm and nucleus. The degradation of ATLs by F-box proteins has not been reported before. In presence of jasmonic acid (JA), which plays a role in anther dehiscence, OsATL53 has been found to gather around the nucleus, and this property enables the translocation of the OsATL53-OsCCR14 complex from a cytoplasmic localization to accumulate around the nuclear periphery. FLIM analyses revealed OsCCR14-OsATL53 complex undergoing conformational changes in presence of JA and this triggers OsFBK1 to mediate the targeted degradation of OsATL53 in the cytoplasm, thereby dissociating the cytoplasmic OsCCR14-OsATL53 complex and enabling OsCCR14 to enter the nucleus and eventually get degraded by SCFOsFBK1 E3 ligase. We have thus studied the signalling mechanism of a variant JA-induced E3 ligase-mediated substrate turnover in plants at the molecular level.


Introduction
Homeostasis in protein levels in all biological processes is a critical and essential regulatory mechanism. The onus of this regulation falls vastly on the ubiquitin 26S proteasome-mediated system (UPS) that is conserved across species. The ligases in the UPS participate in a range of concerted processes including substrate identi cation to mediating controlled ubiquitination of target proteins. The speci city in target protein recognition is particularly expanded in plants where there is a presence of an enormous number of E3 ligases as compared to animals (Jain et al., 2007. Plant E3 ligase genes vastly outnumber those of the other non-plant eukaryotes, where approximately 1400 E3 ligase encoding genes (RING-H2) are the most common RING domains where RING-HC domain has been found in single polypeptide E3 ligases and RING-H2 domain is associated with complex multi-subunit E3 ligases (Jackson et al., 2000). Apart from these domains, the presence of less frequent domains, like RING-v, RING-C2, RING-D, RING-S/T and RING-G type, has also been described (Jiménez-López et al., 2018b). A staggering 551 potential E3 ligases have been reported in Arabidopsis thaliana where 451 correspond to RING-H2 variant and the proportion of this variant in species like Brassica rapa and Oryza sativa are 55% and 57%, respectively (Lim et al., 2010;Alam et al., 2017). It has been previously reported that some RING zinc nger protein genes are rapidly induced in response to elicitors and these genes are members of a family known as ATL (Arabidopsis Tóxicos en Levadura) (Salinas-Mondragón, Garcidueñas-Piña and Guzmán, 1999). These ATLs contain a variant RING-H2 in which the fth cysteine residue has been substituted by a histidine residue. Mutants of ATLs have been previously isolated and analysed to reveal a link between these putative ubiquitin (Ub) ligases and plant defence signalling pathways (Serrano and Guzmán, 2004). It was also reported that the Arabidopsis family consisted of 80 members and rice had 121 ATL genes (Serrano et al., 2006). In the past years, two new classes of RING E3 ligases have been identi ed in addition to ATLs, based on RING-H2 domain variations and RING-outside conserved motifs: BCA 2 Zinc-Finger (BZF) ATLs (BTLs) (Guzmán, 2014) and CTLs containing the YEELL conserved motif (Jiménez-López et al., 2018a).
In the present study, we have investigated the rice ATL gene OsATL53 that is a component of E3 ligase and interacts with OsCCR14 (OsCCR20 (Kawasaki et al., 2006)), the substrate of the F-box protein, OsFBK1. These proteins, OsCCR14 and OsATL53, were identi ed as putative substrates of OsFBK1 in a Y2H screen using anther speci c library, and the degradation of OsCCR14 by SCF OsFBK1 has been described earlier that ultimately regulates ligni cation in rice anthers and roots (Borah and Khurana, 2018). We found that under the in uence of JA, OsATL53 translocates the OsATL53-OsCCR14 complex from the cytosol towards the nucleus where these two substrates are degraded via OsFBK1 in different cellular compartments. Further, we have also shown that OsATL53 affects the enzymatic activity of OsCCR14. We have, therefore, tried to explore the degradation mechanism of protein factors (involved in anther development) in response to hormones like jasmonate, as well as provide a new insight into the function of RING-H2 ATL proteins.

Results
In our previous study (Borah and Khurana, 2018), we had identi ed a few putative substrates of SCF OsFBK1 E3 ligase by a Y2H anther library screening. The cinnamoyl-CoA reductase enzyme, OsCCR14 (previously designated as OsCCR20), involved in catalysing the second rate-limiting step in ligni cation, was found to be a substrate of SCF OsFBK1 ; it was also shown to regulate rice anther ligni cation. The RING-H2 protein, OsATL53, was also identi ed as a putative substrate in the same library screening experiment. Our endeavour, therefore, was to investigate the role of OsATL53 with the aim to understand the molecular mechanism regulating its turnover and its possible relation to rice anther development. The physiological effects of such a mechanism are being currently investigated.

OsATL53 is a component of E3 ligase and interacts with both OsFBK1 and OsCCR14
OsATL53 (RING-H2 protein) is one of the 121 odd RING-H2 ATLs found in rice (Serrano et al., 2006). The signature sequence of ATLs consists of residues that localise to the central region of the RING-H2 domain, viz. two cysteines corresponding to the third and sixth Zn ligands, two histidines occupying the fourth and fth Zn ligands, a highly conserved proline amino acid that separates a residue upstream from the third Zn ligand and a highly conserved tryptophan that is placed three residues downstream from the sixth zinc ligand (Serrano et al., 2006); this also holds true for OsATL53 (PxCxHxxHxxCxxxW, Fig. S1A). Sequence alignment between OsATL53 and known RBX1 proteins studied in humans, yeast and Arabidopsis has also shown that there is homology amongst these proteins (Fig. S1A), even though OsATL53 is not the direct homologue of the canonical RBX1 in reverse blast hits. ATL proteins are also known to comprise of an N-terminal region that is rich in hydrophobic amino acids, and the module generally comprises of a single hydrophobic region of at least 18 residues (Serrano et al., 2006). OsATL53 also has been found to contain one N-terminal hydrophobic transmembrane domain of 22 amino acid residues (Fig. S1B).
In order to con rm whether OsATL53 could be a component of SCF E3 ligase, in vitro immunoprecipitation assay was carried out between GST-CUL5g ( (Borah and Khurana, 2018) and 6xHis-OsATL53, and a weak interaction was detected in this assay (Fig. 1A). We had previously shown that the SCF OsFBK1 E3 ligase complex comprises of OSK1/20, CUL5g (non-standard nomenclature) and OsFBK1 employing Y2H assay (Borah and Khurana, 2018). Based on the above data obtained, it appears that the complete SCF OsFBK1 complex also harbours OsATL53, probably as the RBX1 component (Fig. S1D), although it cannot be ruled out that OsATL53 can be a part of another RING-based E3 ligase.
A detailed study on the architecture of ATLs has suggested that there are two distinct regions in ATLs that are responsible for mediating protein-protein interactions; region I located at the amino-terminal end and region VII that is positioned towards the carboxy-terminal end (Aguilar-Hernández, Aguilar-Henonin and Guzmán, 2011). These two regions have been thought to be important in conferring the functional diversity to ATLs where they might be involved in either target recognition or in mediating the interaction with the components of the functional ubiquitin ligase (Aguilar-Hernández, Aguilar-Henonin and Guzmán, 2011). To test this hypothesis with our RING-H2 protein, we generated GST-tagged deletion constructs that had a truncated N-terminal, ΔOsATL53 (region I, including the transmembrane domain, amino acids from 1 to 32) and a truncated C-terminal end, OsATL53Δ (region VII, amino acids from 138 to 202), respectively ( Fig. S1B for the graphical representation of the deletion constructs).
ATLs like DAF (DEFECTIVE IN ANTHER DEHISCENCE1-Activating Factor) have been previously described to have auto-ubiquitination activity (Peng et al., 2013). First, we investigated the auto-ubiquitination activity of OsATL53 using a protocol described earlier (Peng et al., 2013), with slight modi cations. As is evident from Fig. 1B, OsATL53 has auto-ubiquitination activity and can probably function as an E3 ligase, while the truncated constructs functioning as negative controls showed little or no ubiquitination activity. ΔOsATL53 failed to display any auto-ubiquitination property, while OsATL53Δ showed diminished activity. These results also show that region I of OsATL53 is involved in ubiquitination activity. We also observed a slight shift in size without an apparent smear in the ubiquitinated form of OsATL53 (lane 4 of Fig. 1B) indicating that the addition of ubiquitin is probably in the mono form rather than polyubiquitination.
In general, RING-H2 proteins have been found to localise to different cellular compartments (Furukawa et al., 2000;Greve et al., 2003) and OsATL53 has been found to speci cally localize to the cytoplasm of onion peel epidermal cells ( Fig. 1C and Fig. S3). The red boxes around the nucleus of the target cell in the top (z-stacked cell) and middle (nuclear plane only) panel of Fig. 1C show the absence of EYFP uorescence of OsATL53 in the nucleus while being abundant in the cytoplasm. For all our intracellular localisation experiments using confocal microscopy, we chose to work with onion epidermal cells as it is a monocot system, has a visibly distinguishable nucleus and is a standardized and widely used transient expression system in comparison to epidermal cells of Nicotiana (dicot), and rice protoplasts where the presence of a large vacuole may hamper the viewing of the localization dynamics. As stated earlier, OsATL53 was originally identi ed as one of the putative substrates of OsFBK1 in the Y2H anther library screening along with OsCCR14 (Borah and Khurana, 2018). This interaction was further con rmed by in vitro immunoprecipitation assay between GST-OsFBK1 and 6xHis-OsATL53 (Fig. 1D) and by split EYFP BiFC (Fig. 1E). However, the split EYFP interaction was found to be short-lived and required short incubation time after bombardment (10-12 h), indicating a labile interaction between the two proteins. While OsATL53 is localised solely to the cytoplasm, the split EYFP BiFC between OsFBK1 and OsATL53 showed a dominant cytoplasmic interaction along with a slight presence of the complex in the nucleus (nuclear plane image, Fig. 1E); implying that the interaction between these proteins takes place in both cellular compartments, albeit with an overall cytoplasmic preference.
In a serendipitous experiment, OsATL53 was found to interact with OsCCR14 in a split EYFP BiFC experiment where the EYFP uorescence was found scattered throughout the cytoplasm of the onion peel cells (Fig. 1E). This BiFC interaction was found to be stable even after 16 h post incubation after particle bombardment. This scattered appearance of uorescence, however, did not match with any organellar localisation like that of mitochondria (data not shown). This interaction was further con rmed by in vitro immunoprecipitation (Fig. 1F). However, OsATL53 does not interact with OsCCR18 (OsCCR19 (Kawasaki et al., 2006), Fig. S1E), the closest homologue of OsCCR14 described earlier (Borah and Khurana, 2018), indicating that the OsATL53-OsCCR14 interaction is speci c.
In vitro immunoprecipitation assays were then carried out with OsFBK1 and OsCCR14 separately using both the deletion constructs: ΔOsATL53 and OsATL53Δ. As is evident from the western blot analyses, ΔOsATL53 interacts weakly with OsFBK1 while its interaction with OsCCR14 is unaffected, indicating that region I is involved in interacting with the components of the functional ubiquitin ligase. Whereas for OsATL53Δ, the interaction with OsCCR14 (or the target of SCF OsFBK1 ) is vastly hampered, indicating that the C-terminal domain of OsATL53 is involved in target recognition as reported earlier (Aguilar-Hernández, Aguilar-Henonin and Guzmán, 2011). These experiments also strengthen the idea that RING-H2 proteins might have multiple roles apart from their property to function in the ubiquitination process per se. Putative models of OsATL53-OsCCR14 and OsFBK1-OsATL53-OsCCR14 interactions have also been generated by ab initio modelling (Fig. S2), to gain a perspective into the possible complex formation between these proteins. The models were checked by Ramachandran plot analyses.
OsATL53 is a substrate of OsFBK1 and is preferentially degraded in the cytoplasm Since OsATL53 was identi ed as an interacting protein in the anther library Y2H screening using OsFBK1 as the bait (Borah and Khurana, 2018), we checked whether it is degraded by SCF OsFBK1 . By carrying out in vitro cell-free degradation assays using the total plant protein extracts of 7-day-old seedlings of wildtype (WT), vector control (VC) and OsFBK1 transgenics, as described earlier (Borah and Khurana, 2018), it became evident that OsATL53 degradation is mediated by OsFBK1 via the 26S proteasome pathway ( Fig.  2A and 2B).
We had earlier observed that the BiFC EYFP uorescence showing interaction between OsFBK1-OsCCR14 was preferentially visible in the nuclei of onion peel cells, with weak signals detectable in the cytoplasm ( Fig. 2C; see also (Borah and Khurana, 2018)), while that of OsFBK1-OsATL53 was detected preferentially in the cytoplasm with a weak signal visible in the nucleus ( Fig. 1E and 2E). Such interactions in speci c cellular compartments could provide hint at the target proteins' sites of degradation mediated by SCF OsFBK1 E3 ligase. To prove our assumption, we carried out separate cell-free degradation assays of 6xHis-OsCCR14 and 6xHis-OsATL53 in both nuclear and cytoplasmic protein extracts of OsFBK1 OE and WT lines. The data presented in Fig. 2C and 2D show that degradation of OsCCR14 takes place speci cally in the presence of nuclear protein-enriched extracts under the given experimental duration; while Fig. 2E and 2F show that OsATL53 is degraded in both the protein fractions; it is degraded preferentially in the cytoplasmic protein fraction in comparison to the nuclear protein-enriched extracts within the same assay duration. Curiously, the degradation kinetics for both OsCCR14 and OsATL53 in the respective dominant enriched protein fractions is marginally faster than that observed for total plant protein extracts in the same experimental duration. These experiments provide evidence that OsCCR14, as a substrate of SCF OsFBK1 , is degraded preferentially in the nuclear protein extracts; while OsATL53 is degraded by SCF OsFBK1 in both the nuclear protein-enriched and cytoplasmic protein fractions, with a preference for the cytoplasmic protein extracts. Hence, the BiFC uorescence signals observed for both OsCCR14 and OsATL53 with OsFBK1 indeed indicate the sites of interaction where their subsequent degradation takes place. The labile split-EYFP signals observed for OsFBK1-OsATL53 interaction (Fig. 1E) also correspond to the faster degradation kinetics of OsATL53 mediated by SCF OsFBK1 . That cytosolic enzymes like OsCCR14 are degraded speci cally in the nucleus and mediated by F-box proteins nd support from some earlier studies in animal systems (D'Angiolella et al., 2012).

Jasmonic acid changes the dynamics of the OsCCR14-OsATL53 complex
The process of anther dehiscence preceding pollination is characterised by three distinct events: ligni cation of the secondary walls in endothecial cells, degradation of the septum cells to form a bilocular anther, and the release of pollen grains by the breakage of the septum (Goldberg, Beals and Sanders, 1993). In this process of anthesis, auxin has been shown to regulate the commencement of anther dehiscence by controlling the timing of the ligni cation process at the start of late stamen development (Cecchetti et al., 2008). Various studies have shown that jasmonic acid (JA) is also involved in the control of anther dehiscence mainly in the late stamen development stages (Xie et al., 1998;Stintzi and Browse, 2000;Ishiguro et al., 2001;Devoto et al., 2002). JA is thought to initiate anther dehydration by controlling the transport of water in the anther lament, which in turn triggers stomium breakage and opening at the nal stages of anther development (Sanders et al., 2000;Ishiguro et al., 2001). Thus, it has been suggested that an auxin maximum at early oral stages in Arabidopsis blocks premature endothecium ligni cation, and its decline in the later stages triggers endothecium ligni cation. This reduction in the auxin concentration also leads to the increase in JA concentration, thereby causing stomium breakage and anther dehiscence (Cecchetti et al., 2013).
We had previously shown that process of degradation of OsCCR14 (the enzyme catalysing the second rate-limiting step in lignin formation) by SCF OsFBK1 is not affected by auxin, but transcription of OsCCR14 is induced by auxin (Borah and Khurana, 2018). However, keeping in mind that JA affects the process of anthesis, JA might regulate the process of OsCCR14 degradation to affect anther ligni cation. Based on this assumption, we rst carried out particle bombardment on onion peel epidermal cells of the various constructs (Fig. S3) and treated the peels with 6.4 mM JA for 4 h to check whether JA affects the localization dynamics of these proteins (Fig. S3). The concentration of JA used in our assay (6.4 mM) was as deployed in an earlier study (Cecchetti et al., 2013) that triggered precocious anthesis in Arabidopsis faster than that observed earlier (Stintzi and Browse, 2000). Since 6.4 mM JA is considerably higher than normal physiological conditions, we rst tested the effect of this concentration of JA in onion epidermal cells bombarded by various constructs. We found that the localization patterns of OsFBK1, OsCCR14 and the vector controls were found to be unaffected by JA, even after 4 h of incubation. Fboxes are known to function as hormone receptors (3), but since the localization of OsFBK1 was unaffected after the application of JA to the bombarded cells (Fig. S3), we can presume that OsFBK1 does not function as a JA sensor. However, the EYFP uorescence of OsATL53 appeared to accumulate around the nucleus but did not enter the nucleus (Fig. S3 and Fig. S4; for cells with nuclear stain). This accumulation of OsATL53 around the nucleus was seen to initiate as early as 2 h post incubation (data not shown). In order to rule out the effect of stress caused by a high JA concentration in the bombarded cells during the experimental time frame, we checked the changes in lifetime of EYFP-OsATL53 on exposure to a 6.4 mM JA by FLIM confocal microscopy (Fig. S5B), as the lifetimes of uorophores are affected by the cellular pH and ionic changes amongst other factors (Trautmann et al., 2013). Even after 4 h of incubation with 6.4 mM JA, the lifetime of EYFP-OsATL53 did not change (Fig. S5C). A representative stressed or plasmolysed cell expressing EYFP-OsATL53 is also shown as a control that displays a decrease in lifetime due to changes in cell integrity ( Fig. S5B and S5C). Further, we also checked the changes in localization of EYFP-OsATL53 with 100 µM JA, the concentration used in experiments done earlier (Stintzi and Browse, 2000). However, changes in localization of EYFP-OsATL53 were observed to initiate only after 8 h of incubation with 100 µM JA as opposed to 2 h when 6.4 mM JA was used (Fig. S4A). Due to the large difference in the time of incubation and the reason that both concentrations do not affect the cellular integrity, we chose to use 6.4 mM JA for all further experiments.
We had also observed that OsATL53 and OsCCR14 interacted with each other, therefore, we carried out co-localization studies using respective constructs and subjected the bombarded onion peel cells to JA treatment for 4 h. As is evident from Fig. 3A, a distinct accumulation of EYFP (OsATL53) and ECFP (OsCCR14) uorescence is observed around the nucleus after 4 h of JA treatment. While the localization of OsCCR14 is unaffected by JA when present alone (Fig. S3), there is a de nite presence of OsCCR14 uorescence along with OsATL53 around the nucleus after JA treatment. This implies that the localization of the complex of OsCCR14-OsATL53 is affected by JA, presumably mediated by the movement of OsATL53 towards the nucleus on exposure to JA (Fig. 3A). On the other hand, colocalization of OsFBK1-OsATL53 and OsFBK1-OsCCR14, respectively, were unaffected even in the presence of JA (Fig. S6). However, in both the absence and presence of JA, OsATL53 was found to be present slightly in the nucleus when co-localized with OsFBK1 (Fig. S6). This observation for OsCCR14-OsATL53 complex localization in response to/without JA was also repeated using BiFC interaction studies (Fig. 3B, OsCCR14-nEYFPC 1 and OsATL53-cEYFPN 1 ). However, we failed to observe any EYFP uorescence of the BiFC interaction beyond 4 h of JA exposure.
Additionally, we also investigated whether, apart from the movement of the OsCCR14-OsATL53 complex and its accumulation around the nuclear periphery, JA also induces structural changes in this complex. For this we adopted the FLIM-FRET (Fluorescence Lifetime Imaging-FÖrster Resonance Energy Transfer) technique to examine any changes in the lifetime of the ECFP-donor construct with/without JA. This technique of FLIM-FRET using the latest Leica SP8 Falcon confocal microscope has been recently used to study the essential role played by the Zn+ transporter ZIP7 in B cell development in mice (Anzilotti et al., 2019). Since we already know by BiFC analysis that OsCCR14 and OsATL53 interact in a particular vector combination that allows the split EYFP to come close together for uorescence, thus, we used FRET probes that have both the uorescent probes (ECFP and EYFP) in the N-terminal regions, and technically will be separated beyond the required FÖrster distance (<10 nm). As expected, the lifetime of the donor ECFP (~3 ns in solution, https://www.fpbase.org/, (Lambert, 2019)) did not change under FLIM-FRET condition (Fig. 3C and 3D). However, when we added JA and incubated the bombarded onion peel cells for 2 hours, we noticed that the lifetime of the donor ECFP had decreased in FLIM-FRET conditions indicating a positive FRET (Fig. 3C and 3D). Positive FRET was not only observed around the nucleus where the ECFP-OsCCR14 and EYFP-OsATL53 uorescence were found to accumulate, but also throughout the cell (Fig. 3D). This meant that the addition of JA induces a certain conformational change to the OsCCR14-OsATL53 complex that enables the two FRET probes to come within the minimum FÖrster distance and cause energy transfer. Such a change in donor lifetime was not observed in the control (OsFBK1-EGFP donor, EYFP-OsCCR14 acceptor) (Fig. S7). Such conformational changes might also explain the reasons behind the non-observation of BiFC EYFP uorescence between OsCCR14-OsATL53 after 4 h of JA exposure (Fig. 3).
We had also mentioned that the BiFC interaction between OsCCR14-OsATL53 was stable for long duration post incubation after particle bombardment. However, JA affected the localization of the split-EYFP interaction between these two proteins in the onion epidermal peel cells. Therefore, we wanted to investigate whether JA plays a role in maintaining the stability of the OsCCR14-OsATL53 complex in rice. To explore this possibility, we carried out cell-free degradation using bacterially expressed 6xHis-OsCCR14 and 6xHis-OsATL53 puri ed protein fractions and incubating them with the whole protein extracts of WT rice seedlings as described previously (Borah and Khurana, 2018). It is clear from Figure 3C that JA affects the stability of OsCCR14-OsATL53 complex after 4 h of treatment by hastening the degradation of OsATL53, without affecting the stability of OsCCR14 during the experimental timeframe. However, in the absence of JA, OsATL53 has been found to be stable in the cell-free experiments when present as a complex with OsCCR14 (Fig. 3E). This is a stark contrast to the observation where the half-life of OsATL53 has been found to be considerably less when present alone in the cell-free experiments ( Fig.  2A). The results of the cell-free experiments have been quanti ed and graphically represented in Fig. 3F. These observations lead us to speculate that OsATL53 might interact with OsCCR14 to form a complex and probably attenuates the enzymatic reactions carried out by OsCCR14 in the ligni cation process and is turned over by SCF OsFBK1 when present alone.

Jasmonic acid induces changes in the OsFBK1/OsCCR14-OsATL53 interaction and localization dynamics
It is apparent from the data presented above that OsFBK1 interacts with both OsCCR14 and OsATL53 individually to mediate their degradation via the 26S proteasome pathway in different cellular components. We had also observed that OsATL53 probably enters the nucleus while forming a complex with OsFBK1 ( Fig. 1E and Fig. S6). Keeping in mind these observations, it was imperative to check the nature of interaction between OsFBK1 and OsCCR14-OsATL53 complex and determine whether JA affects interaction amongst all these proteins. For this, we carried out particle bombardment for a threeway interaction amongst ECFP-OsFBK1 and OsCCR14-OsATL53 split EYFP. The upper panel in Fig. 4 shows that under normal untreated conditions, OsFBK1 does not seem to overly interact with OsCCR14-OsATL53 complex and follows its normal localization pattern, although some minor interactions could also be seen (Fig. 4A, upper panel). However, when the bombarded onion epidermal peels were treated with JA (+MG132) and incubated for 4 h, a drastic change in the dynamics of the interactions amongst the three proteins was observed (lower panel, Fig. 4A). The green colour shows the overlapping regions between the ECFP and EYFP uorescence and it also shows that OsFBK1 interacted with OsCCR14-OsATL53 complex while accumulating around the nucleus, as observed earlier for JA treated OsCCR14-OsATL53 EYFP uorescence. At the same time, in most cells we could also see that OsFBK1 has migrated out of the nucleus into the cytoplasm to interact with the OsCCR14-OsATL53 complex ( Fig. 4A and B). Curiously, we did not observe any such behaviour for OsFBK1 alone when treated with JA (Fig. S3).
Further, we also observed that on exposure to JA alone (-MG132), there was a loss of EYFP signal in the target cells indicating the disruption of the OsCCR14-OsATL53 complex and the possible degradation of the components in the respective compartments, as also proved by the cell-free studies (Fig. 4B, Fig. S8). At the same time, the localization pattern of OsFBK1 also seemed to change back to its original state after this process of degradation is over and its uorescence appeared throughout the cell and in the nucleus (Fig. 4B). We had also noticed that some target cells exhibited faster movement and degradation kinetics, some initiating as sooner as 40 min post JA treatment (Fig. 4B) in the absence of MG132 as compared to the cell-free studies ( Fig. 3E and F). This is expected as the cell-free studies had considerably higher substrate concentration (5 µg) than available in native cellular conditions.

OsATL53 attenuates enzymatic activity of OsCCR14
Finally, since OsATL53 forms a stable complex with OsCCR14 in the cytoplasm, we investigated whether it affects the enzymatic activity of OsCCR14. OsCCR14 is a cytosolic enzyme and has been found to preferentially reduce feruloyl-CoA esters into coniferaldehyde in the cytoplasm (Park et al., 2017). Similarly, SbCCR1 (the sorghum ortholog of OsCCR14) also has a higher a nity for feruloyl-CoA ester as a substrate as opposed to the other esters (p-coumaroyl CoA, sinapoyl-CoA) (Sattler et al., 2017). Thus, we chose feruloyl-CoA as a substrate for investigating the enzymatic activity of OsCCR14 alone, and in complex with OsATL53 (with/without JA). Feruloyl-CoA ester was prepared following the protocol described earlier (Park et al., 2017;Sattler et al., 2017) and used as substrate for the reaction mix comprising of: a) OsCCR14, b) OsCCR14 + OsATL53, c) OsATL53, d) OsCCR14 + OsATL53 + JA, and e) OsCCR14 + JA.
As evident from Fig. 5A, OsCCR14 shows a K m of 6 µM, whereas the formation of a complex with OsATL53 demonstrates a signi cant reduction in the substrate a nity of OsCCR14 (K m 20 µM). The addition of JA to either OsCCR14 or to the complex, however, does not affect the K m values of OsCCR14 and the OsCCR14-OsATL53 complex, respectively, but slightly affects both their K cat values (Fig. 5B), thereby signifying a marginal decrease in enzyme activity.
Our data clearly show that OsATL53 attenuates the enzymatic activity of OsCCR14 by signi cantly reducing its a nity for its desired substrate, feruloyl-CoA, without totally inhibiting its enzymatic activity. We might speculate at this juncture that OsATL53 acts as an allosteric effector molecule rather than an inhibitor. However, it would require detailed biochemical experimentation to con rm this assumption. Additionally, the presence of JA does not reduce the substrate a nity of OsCCR14 as a complex with OsATL53 any further, but the changes brought about in the conformation of the complex in its presence (as shown earlier, Fig. 3) is a trigger for the translocation of the complex towards the nucleus, and for OsFBK1 to recognise the complex as a substrate to initiate its degradation via the 26S proteasome pathway (Fig. 4).
However, based on our data in its current form, we cannot comment whether the ternary complex of OsFBK1-OsATL53-OsCCR14 enters the nucleus as a unit or whether they enter as separate binary complexes after partial degradation of OsATL53. We intend to pursue the molecular and biochemical factors involved in this JA induced changes in the OsFBK1/OsCCR14-OsATL53 interaction and localization dynamics. Additionally, the effects of such interaction between OsATL53 and OsCCR14 at the physiological level is currently being worked upon by the generation of knock-down rice transgenics. Our data also strongly indicate that there might be other protein partners at play or the presence of a unique signalling mechanism that remains elusive presently.

Discussion
The ATL family of RING-H2 proteins are a dynamic group that has been studied in the recent years. Several ATL proteins have been explored in Arabidopsis and other plant species and diverse functions have been attributed to them (Guzmán, 2012). The Arabidopsis and rice ATL family of RING-H2 proteins have been phylogenetically distributed into several clades based on their architecture and this large family of proteins are thought to be functioning as putative ubiquitin ligases of the RING-H2 type (Serrano et al., 2006). The diversity in the architecture of the ATLs has been proposed to recognise and target substrates in different cellular environments (Aguilar-Hernández, Aguilar-Henonin and Guzmán, 2011). The ATL in our study was previously identi ed as OsATL53 and assigned to clade 'k' and is one of the few ATLs that is encoded by a gene containing introns (Serrano et al., 2006). We found OsATL53 to be a component of a putative SCF OsFBK1 and to have auto-ubiquitination properties (Fig. 1). Based on the results obtained by our previous Y2H anther library screening experiments (Borah and Khurana, 2018), we con rmed that OsATL53 interacts with the F-box protein OsFBK1 and also with the cinnamoyl CoAreductase enzyme OsCCR14, which was also a substrate of OsFBK1 (Fig. 1). However, the cellular localization of these two interactions differed indicating that OsATL53 interacts with these two proteins differently. These variations in the interactions could be due to the different domains present in OsATL53 as described earlier (Aguilar-Hernández, Aguilar-Henonin and Guzmán, 2011) and have been proved true by deletion studies (Fig. 1).
We found that OsFBK1 degrades OsATL53 and this degradation is pronounced in the cytoplasmic extract as compared to nuclear extract, which is in stark contrast to the degradation dynamics of OsCCR14 (Fig.   2). Seeing that both these proteins are substrates of SCF OsFBK1 and, OsATL53 itself is likely to be a component of the same E3 ligase, it appears that OsATL53 performs other functions apart from participating in the ubiquitination process. However, RING-H2 or ATLs being recognised as substrates by F-box proteins and degraded by them have not reported before. OsCCR14 has been found to be the closest homologue to AtCCR1 and has been suggested to be primarily involved in actively lignifying tissues like stems and roots (Park et al., 2017). It was also found that OsCCR14 expresses constitutively throughout the various developmental stages of rice (Park et al., 2017). Since OsCCR14 and OsATL53 were both identi ed as single target clones in the anther library, we speculate that these two proteins might be together involved in the ligni cation process and, from the rice anther development point-ofview, OsATL53 might also be involved in the degradation process of OsCCR14 after anther ligni cation process is over. Jasmonic acid has been previously studied extensively for its role in anther dehiscence (Sanders et al., 2000;Ishiguro et al., 2001;Cecchetti et al., 2013). Additionally, the ubiquitin system and jasmonate signalling pathway are closely linked where the ubiquitin system contributes to the regulation of the synthesis, perception and signal transduction of JA (Nagels Durand, Pauwels and Goossens, 2016). Keeping these facts in mind, we investigated the effect of JA on the interactions between OsATL53-OsFBK1 and OsCCR14-OsATL53. While the interaction between OsATL53-OsFBK1 was unaffected by JA, OsCCR14-OsATL53 complex was found to accumulate around the nucleus when treated with JA (Fig. 3). This movement towards the nucleus after JA treatment has been observed for OsATL53 alone also (Fig. S3), and thus the accumulation of the OsCCR14-OsATL53 complex around the nucleus could be attributed to OsATL53.
Although, OsCCR14 can enter the nucleus on its own, it is not capable of transporting OsATL53 into the nucleus (Fig. 3). However, this is not the case for the OsFBK1-OsATL53 complex where OsATL53 seems to have entered the nucleus with the 'assistance' of OsFBK1 (Fig. S6) and this process seems to be unaffected by JA. We had seen that under the in uence of JA, the erstwhile stable OsCCR14-OsATL53 complex was translocated around the nucleus, and that OsATL53 is degraded both in the cytoplasm and the nucleus and OsCCR14 is degraded solely in the nucleus. We had also seen that JA induces conformational changes in the OsCCR14-OsATL53 complex (Fig. 3C-D, Fig. S7). Based on these observations, we assumed that JA somehow disrupts the stability of the OsCCR14-OsATL53 complex, thereby enabling the targeted destruction of both the components by SCF OsFBK1 in their respective cellular compartments. Fig. 4B and Fig. S8 clearly show that JA indeed destabilises the OsCCR14-OsATL53 complex in the cytoplasm and results in the rapid degradation of OsATL53. This was further shown in the cell-free studies too (Fig. 2). Incredibly, OsATL53 was found to be stable as a complex with OsCCR14 in the absence of JA. RBX1 RING-H2 proteins have been quoted to "function as allosteric activators of E2 enzymes and play an essential role in ubiquitination process, although their precise mechanism of action is unknown" (Freemont, 2000;Jackson et al., 2000;Noureddine et al., 2002), which made us wonder whether plant ATLs might also participate in other enzymatic processes apart from ubiquitination of target substrates. Thus, we provide evidence that the RING-H2/ATL protein, OsATL53, interacts with OsCCR14 and might function to decrease its enzymatic activity.
Further, we still needed to account for the role of OsFBK1 in this JA-induced translocation of OsCCR14-OsATL53 complex. We found that in an uninduced state, OsFBK1 does not interact with the OsCCR14-OsATL53 complex even though it interacts with the proteins individually (Fig. 4A). But, as soon as JA is applied, we see a change in the localization of OsFBK1 in the cell, whereby OsFBK1 not only interacts with the OsCCR14-OsATL53 complex in the cytoplasm but also travels with the complex towards the nucleus (albeit not appearing to enter the nucleus with the OsCCR14-OsATL53 complex) (Fig. 4A). At the same time, we had also shown that OsFBK1 mediates the degradation of OsATL53 in the cytoplasm ( Fig.   2E and 3E). Based on these data, we believe that during this time, after JA exposure, OsFBK1 starts to mediate the degradation of OsATL53 in the cytoplasm, thereby destabilising the OsCCR14-OsATL53 complex and enabling OsCCR14 and the remaining OsATL53 to enter the nucleus to be degraded via the 26S proteasome and mediated by SCF OsFBK1 . We also believe that the JA-induced conformational changes brought about in the OsCCR14-OsATL53 complex ( Fig. 3C and D) might be the trigger for the movement and subsequent degradation of the complex by OsFBK1, even though we are not aware of the factors involved in such a phenomenon at the molecular level at present. Such conformational changes brought about by JA in proteins and their subsequent degradation by the 26S proteasome have been previously reported for the JAZ degron (Sheard et al., 2010).
From the biological point-of-view and to determine whether OsATL53 has a role in regulating the functions of OsCCR14, we examined the dynamics of enzymatic activities of OsCCR14 both alone and as a complex with OsATL53, in the presence and absence of JA, respectively. Since OsCCR14 catalyses the rst limiting step in lignin formation in the cell cytoplasm, our data reveals that OsATL53 slows down the enzymatic processes of OsCCR14 and, thereby, functions to regulate ligni cation. Microarray data obtained from previously conducted experiments (Jain et al., 2007) have shown that the expressions of OsCCR14 and OsATL53 overlap in tissues like roots of 7-day-old seedlings, in orescence stages like P4 and P6 and seed stage S5, whereas their expression differs in the other stages of development examined (Fig. S8). Based on these data, it can be suggested that OsATL53 affects the ligni cation processes catalysed by OsCCR14 in different target tissues depending on the requirements of that development stage of rice plant. The presence of JA in tissues like late anther stages, however, changes the dynamics of the enzymatic processes and signals the recognition of the OsATL53-OsCCR14 as substrates for destruction via the SCF OsFBK1 E3 ligase.
Finally, to explain all the evidences we generated coherently in context to rice anther development (as both OsCCR14 and OsATL53 were screened in yeast anther library), we propose the following model ( Fig.  6): During the initial anther development stages where ligni cation is controlled by auxin, OsCCR14 executes its enzymatic processes in the cytoplasm whereas the increased presence of OsATL53 in the later stages reduces this enzymatic activity. However, towards the later stages of anther maturation preceding dehiscence, when the level of JA starts to increase, the process of ligni cation needs to cease, and the proteins involved need to be turned over. At this stage, the presence of increasing levels of JA in the cell might change the conformation of the OsCCR14-OsATL53 complex thereby triggering OsATL53 to travel towards the nucleus (via as yet unknown mechanism), while carrying OsCCR14 along with it as a payload. Although OsFBK1 previously did not interact with the complex but this JA-induced movement of the OsCCR14-OsATL53 complex towards the nucleus as substrates is recognised by OsFBK1 (mechanism remains unknown). At the same time, the substrate(s) bound functional SCF OsFBK1 E3 ligase might initiate the degradation of OsATL53, thereby disassociating the OsCCR14-OsATL53 complex. This would in turn enable OsCCR14 to enter the nucleus where it is degraded by the 26S proteasome. Also, some of the remaining OsATL53 might enter the nucleus as an OsFBK1-OsATL53 complex where OsATL53 would also be nally degraded via the nuclear 26S proteasome, mediated by the SCF OsFBK1 E3 ligase.
To summarise, we have reported a divergent role of ATL RING-H2 proteins in interacting with other proteins, regulating their activities, and in turn become substrates of E3 ligases, thereby highlighting the diversity in their functions apart from participating as a bridging partner between the E3 and E2 ligases. Additionally, the modulation of OsCCR14 enzymatic function by OsATL53 could have application in terms of manipulating the lignin content in cash crops. At the same time, we could also gain an insight into the functioning of the F-box protein, OsFBK1, in targeted substrate degradation that would in turn regulate anther development in rice with precision. We have also tried to highlight a possible pathway of substrate degradation that is triggered by jasmonate.
Additionally, OsFBK1 was cloned in the pSITE-1CA vector for CFP uorescence by Gateway TM cloning.
The ampli cation of the genes was carried out using Phusion TM high-delity Taq polymerase (Finnzymes, Finland) as per manufacturer's instructions (see Supplemental Table S1 for primers).

Particle bombardment
Biolistics of the various constructs on onion epidermal peel cells for BiFC, intercellular localization and hormone assays were carried out using Biolistic PDS-1000/He particle delivery system (Bio-Rad, USA) as described earlier (Borah and Khurana, 2018). For hormone-related visualisation, the bombarded and incubated onion peels (12-16 h) were treated with 6.4 mM JA for the requisite time-points before observing for uorescence under confocal microscope (Leica TCS, SP5). For FLIM-FRET experiments, all observations and analyses were carried out using the Leica SP8 FALCON confocal microscope at the Central Instrumentation Facility (CIF), University of Delhi South Campus. FLIM was carried out at 40 MHz and curve-tting was carried out using the n-Exponential Reconvolution t model. For all experiments, the tissue was mounted in water and observed at 20X magni cation. All experiments were carried out at least ve times, with an average of 3 cells being visualised in each repetition.

Auto-ubiquitination assay
The auto-ubiquitination activity of OsATL53 was carried out following the protocol described by Peng et al. (2013) with slight modi cations. An E1 and E2 ligase mix (Cat. # U8382, Sigma, USA) was used for the activation and conjugation of ubiquitin. Negative controls used were generated in the laboratory. Probing of ubiquitin was done using monoclonal anti-ubiquitin antibodies (Cat. # U0508-50UL, Sigma, USA) at a dilution of 1:8,000+1% BSA.

Cell-free degradation
For cell-free degradation assays, total plant protein was extracted from 7-day-old seedlings of WT and OsFBK1 transgenics and cell-free experiments were carried out using 5 µg 6xHis-OsATL53/5 µg 6xHis-OsCCR14 in 30 µg total plant protein extracts/nuclear-enriched protein extracts/cytoplasmic protein extracts by following the protocol described before (Borah and Khurana, 2018). Nuclear protein extracts from 10-day-old seedlings (5 g tissue) were obtained as per protocol described by (Xu et al., 2012). The cytoplasmic extracts were aliquoted after ltration and the rst centrifugation step. For loading controls, the same blots were stained with Ponceau for the detection of the Rubisco protein large subunit band for cell-free with total plant protein and cytoplasmic extracts whereas for cell-free with the nuclear-enriched protein extracts, Ponceau stained histone H3 band (~15kDa) was used as loading control. For cell-free experiments with JA, the cell-free degradation cocktail was mixed with 6.4 mM JA and incubated at 30 o C for the requisite time points. The degradation data was graphically represented by measuring the intensities of the bands of all blots in Image Studio TM Lite ver. 5.2 (https://www.licor.com/bio/products/software/image_studio_lite/). The 2 h/6 h western band for each blot of OsATL53 and OsCCR14, respectively, was used as the reference and measured intensities were normalised against it and multiplied by 100. Representation of the data is in percentage intensity ratio. All experiments were carried out at least thrice and error bars denote standard deviation across replicates.

Synthesis of feruloyl-CoA thioester and puri cation
The protocol for synthesis of the substrate feruloyl-CoA thioester was modi ed from the processes described earlier (Park et al., 2017;Sattler et al., 2017). Brie y, the Arabidopsis4-coumarate: CoA ligase 1 gene, At4CL1, was ampli ed from leaf cDNA and cloned in pET28a vector (see Table S1 for primers). The clone was transformed in BL21-RIL competent cells and the protein induced and puri ed as described earlier (Borah and Khurana, 2018). For the synthesis of the thioester, 410 µg of puri ed At4CL1 protein was added to a 40 mL solution containing 50 mM sodium phosphate buffer, pH 7.0, 800 µM CoA (Cat. # C4282, Sigma, USA), 2.5 mM ATP (Cat. # A-7699, Sigma, USA), 2 mM ferulic acid (Cat. # 128708, Sigma, USA) and 5 mM MgCl 2. The reaction mix was incubated at room temperature for 20 h at low rpm in dark conditions. Thermal denaturation of the enzyme was carried out at 80 o C for 10 min, and pelleted at 15,000g for 30 min at room temperature. The supernatant was dried under vacuum centrifugation to collect solutes. The solutes/substrates were puri ed by ethanol for 6 h at -20 o C and the supernatant was dried under vacuum centrifugation. The obtained pellet/substrate was stored at -20 o C for later use. A standard curve of the obtained ester was generated using A 366 for measurement using a spectrophotometer.
For reactions with OsCCR14 and OsATL53 complex, equal amounts of OsCCR14 and OsATL53 were rst incubated at 4 o C for at least 2 h, quanti ed and 3 µg used per reaction. For reactions with JA, 100 µM JA was added to each reaction and incubated at 4 o C for 3 h. Reactions were initiated by the addition of the enzyme for 20 min with an interval of 1 min between reactions (20 reactions for one substrate concentration per reaction condition). Reactions were stopped by the addition of 40 µL of 15% (w/v) trichloroacetic acid and mixed well. These were then centrifuged brie y and 200 µL of each reaction mix was loaded onto a Corning clear 96-well plate. Decrease in A 366 (substrate concentration) was recorded using a Tecan M200 microplate reader (Tecan Austria GmbH) and the i-control 1.7 software (Tecan Austria GmbH). K m , V max and K cat values were determined by extrapolation from Lineweaver-Burk plots.
Each enzyme assay was carried out for ve times and the average values represented the mean ± standard deviation.

Declarations
Funding: The Department of Biotechnology, Government of India (BT/PR12394/AGIII/103/891/2014) provided the nancial support. J.P.K. also acknowledges the award of J.C. Bose National Fellowship by the SERB, Government of India (SB/S2/JCB-13/2013); P.B. availed the Research Associateship sanctioned as part of the DBT project. A.S. would like to thank the University Grants Commission, New Delhi, for the award of a Research Fellowship. For infrastructural support, authors acknowledge the nancial assistance provided by the Department of Science and Technology (through FIST and PURSE Grants), the University Grants Commission (through UGC-SAP grant) and the University of Delhi.