TurboID-based proteomic profiling reveals proxitome of ASK1 and CUL1 of the SCF ubiquitin ligase in plants

Abstract

The SKP1-Cullin-F-box (SCF) complex is one of the best-studied E3 ubiquitin ligases in plants because of its critical roles in various signaling pathways (Lechner et al., 2006; Sadanandom et al., 2012; Stefanowicz et al., 2015; Abd-Hamid et al., 2020). The core components of SCF include Cullin 1 (CUL1) scaffold protein that interacts with SKP1 adaptor protein at the N-terminal, and E2-interacting RING-finger protein RBX1 at the C-terminal (Fig. 1a). The SKP1 interacts with interchangeable F-box receptor units that specifically recognize target substrates for ubiquitination and degradation by the 26S proteasome (Zheng & Shabek, 2017). Among the 21 SKP1-like (ASK) proteins in Arabidopsis, 19 exhibit significant structural similarity and are thought to be functionally redundant. Historically, studies have focused on ASK1 due to its prominent function as an adaptor in the SCF module, with higher steady-state levels of expression throughout the plant, particularly in proliferating tissues, compared with other ASK proteins (Porat et al., 1998; Yang et al., 1999; Zhao et al., 1999; Gagne et al., 2002; Risseeuw et al., 2003). The Arabidopsis genome is predicted to encode hundreds of F-box proteins that target thousands of substrate proteins, but the majority remain uncharacterized and without known substrates (Abd-Hamid et al., 2020). Interestingly, mammalian SKP1 has recently been shown to regulate the switch between autophagy and unconventional secretion (Li et al., 2023), suggesting that its biological function extends beyond acting as an adapter protein within the SCF complex. Thus far, identifying E3 substrates has been challenging because substrates often interact weakly and transiently with the E3, and they are rapidly degraded and hence difficult to capture (Pierce et al., 2009; Iconomou & Saunders, 2016). Although traditional methods such as affinity purification coupled to mass spectrometry (AP-MS), yeast-two hybrid (Y2H), and protein microarray screening have captured E3 substrates with some success, each has its own drawbacks (Harper & Tan, 2012; Iconomou & Saunders, 2016). AP-MS fails to capture weak and transient interactors, whereas Y2H is labor intensive, prone to false positives, and is a heterologous system. The recently developed proximity labeling (PL) approach to capture protein–protein interactions (PPI) in vivo overcomes many of the drawbacks of traditional approaches. For PL, a protein of interest (POI) is fused to a promiscuous biotin ligase. This ligase catalyzes biotin to a short-lived biotinoyl-5′-AMP, which can diffuse away from the ligase and react with amine groups on lysine residues of nearby proteins, typically within a radius of 10 nm (Kim & Roux, 2016; Qin et al., 2021; Yang et al., 2021). The biotinylated proximal interactomes (i.e. proxitomes) of the POI can be enriched using streptavidin-conjugated beads under stringent conditions followed by proteome identification through mass spectrometry (MS; Kim & Roux, 2016; Qin et al., 2021; Yang et al., 2021).

In this study, we utilized the TurboID biotin ligase, which is highly efficient at transferring biotin to proximal proteins in many organisms, including plants (Branon et al., 2018; Mair et al., 2019; Zhang et al., 2019), to identify the SCF interactome in Arabidopsis. We uncovered the CUL1 and ASK1 proxitomes, providing insights into the functions of interactors in both conventional SCF-mediated signaling pathways and potentially noncanonical SCF-independent processes. Further investigation and analysis of the interactomes associated with ASK1 and CUL1 revealed novel partners involved in diverse biological processes within plants. The TurboID-based PL strategy detailed in this study can be expanded to identify targets belonging to other E3 families in plants. This expansion would enhance the repertoire of components within the ubiquitin-proteasome system and shed light on their pivotal roles in regulating a multitude of cellular processes.

In the SCF complex, CUL1 associates with ASK1, and ASK1 recruits F-box proteins along with their target substrates for degradation (Fig. 1a). Therefore, we used Arabidopsis CUL1 and ASK1 as bait to capture SCF targets under normal growth conditions (Fig. 1a). To that end, we fused 3xMyc epitope tag and TurboID to the N-terminus of ASK1 (TurboID-ASK1) under the control of Arabidopsis ubiquitin promoter (pUBQ) and NOS terminator (Fig. 1a, upper panels and Fig. S2a). Similarly, we constructed the NTD of CUL1 fused to TurboID-3xMyc (TurboID-CUL1^NTD). The CUL1^NTD was designed as described previously (Zheng et al., 2002), to enrich and stabilize E3 targets by eliminating the CUL1-RBX1 region and subsequent recruitment of E2 ubiquitin-conjugating enzymes. As a control to account for background biotinylation, we used Citrine fluorescent protein fused to TurboID-3xMyc (Citrine-TurboID; Fig. S2a). These constructs were transformed into Arabidopsis Col-0 plants, and T3 homozygous lines were selected and validated (see the Materials and Methods section for details). At 4 wk, TurboID-CUL1^NTD, TurboID-ASK1, and Citrine-TurboID transgenic lines appear phenotypically similar to wild-type (WT) Col-0 plants (Fig. 1a, lower panels). The expression of TurboID fusion proteins in these lines was confirmed by immunoblot analysis (Fig. S2b–d). Among different methods, we tested for biotin application: syringe infiltration of biotin into leaves delivered biotin more efficiently and caused less tissue damage compared with vacuum infiltration and submersion in the biotin solution. Subsequent time course experiments and immunoblotting with streptavidin-conjugated HRP revealed that 50 μM biotin infiltration and incubation for 3-h postinfiltration at room temperature was sufficient for efficient labeling activity for all tested TurboID fusions (Fig. S2b–d). In WT Col-0 plants, infiltration of 200 μM biotin followed by incubation at 3 h at room temperature resulted in detection of very few biotinylated proteins (Fig. S2b,d, last lane). These results indicate that biotin-labeled proteins were observed exclusively in the transgenic lines expressing TurboID fusions.

To identify ASK1 and CUL1^NTD proximal proteins, we performed affinity purification of biotinylated proteins in three replicates (Fig. 1b; see the Materials and Methods section for details). Immunoblot analysis confirmed the successful enrichment of biotinylated proteins on beads (Fig. S2e). The biotinylated proteins bound to streptavidin beads were eluted using boiling SDS supplemented with 12.5 mM biotin. Eluted proteins were cleaned and digested. The resulting peptides were labeled with distinct TMT and analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS; Fig. 1b). Our analysis identified a total of 2361 proteins and quantified 2241 (Table S2a–c). Biological replicates taken from plants expressing the same transgene showed a high Pearson correlation (c. 0.99; Fig. S3). Among the identified proteins, 705 and 404 were considered ASK1 and CUL1 interactors, respectively, based on the threshold of q-value ≤ 0.1 and fold change (ASK1/Citrine or CUL1/Citrine) ≥ 1.23 (Fig. 1c; Table S2d,e). Comparison between ASK1 and CUL1 interactors identified 120 common interactors (Fig. 1c; Table S3). The unique and common interactors include F-box proteins and other ubiquitin-related proteins (Fig. 1c; Tables S4, S5).

To understand the functional significance of enriched ASK1 and CUL1 interactions, we performed Gene Ontology (GO) BP enrichment analysis. A large number of CUL1 interactors were related to protein metabolic process and gene expression categories (i.e. RNP biogenesis, translation, and RNA splicing; Fig. 1d; Table S6). Conversely, ASK1 interactors were, mostly, spread in smaller groups across various significantly enriched BPs. For both datasets, the BP term ‘SCF-dependent proteasomal ubiquitin-dependent protein catabolic process’ (referred to as ubiquitin-related) was enriched (Fig. 1d; Table S6). To further characterize ASK1/CUL1 interactors from our list, we compiled a list of F-box proteins and proteins related to ubiquitination or proteasomal activity and looked for them in our proxitome dataset. Statistical testing showed significant enrichment of ‘ubiquitin-related’ proteins among ASK1 interactors (P-value = 0.0014), and CUL1 interactors (P-value = 0.0015, hypergeometric test; Table S7).

In the CUL1^NTD interactors, previously described F-boxes were among the most highly enriched. These include AUXIN SIGNALING F-BOX 3 (AFB3), CORONATINE INSENSITIVE 1 (COI1), F-BOX AND LEUCINE RICH REPEAT DOMAINS CONTAINING PROTEIN (FBD), CONSTITUTIVE EXPRESSER OF PR GENES 1 (CPR1), and SLOW MOTION (SLOMO) (Dharmasiri et al., 2005; Xu et al., 2002; Kuroda et al., 2012; Gou et al., 2009; Lohmann et al., 2010; Table S5a). The F-boxes AFB2, AFB5, and FBD were most highly enriched in the ASK1 data set, as well as PHYTOCHROME A (PhyA), MAPK/ERK KINASE KINASE 1 (MEKK1), PIN-FORMED 1 (PIN1), EXOCYST SUBUNIT EXO70 FAMILY PROTEIN A1 (EXO70A1), MITOGEN-ACTIVATED PROTEIN KINASE KINASE 5 (MKK5), and OLEOSIN 1 (OLE1) (Table S5b), which have been previously demonstrated to undergo 26S proteasome-mediated degradation but were not specifically shown to be targeted by SCF E3 complex (Clough & Vierstra, 1997; Dharmasiri et al., 2005; Nakagami et al., 2006; Nguyen et al., 2013; Traver & Bartel, 2023).

Given that 77 proteins in our dataset are explicitly grouped under ubiquitin-related proteins and our dataset also included 12 previously described interactors of ASK1 and CUL1 (Fig. 1c; Table S5), we postulated that our ASK1 and CUL1^NTD TurboID fusions captured a mixture of SCF substrates and F-boxes, most of which are novel. To test whether these are bona fide target substrates of the SCF E3 ligase, we selected a dozen proteins for further in planta validation (Table S8). First, we performed a BiFC assay to confirm the interactions using Agrobacterium-mediated transient expression in N. benthamiana followed by confocal microscopy. For this, we co-expressed candidate interactors fused to the N-terminal 155 amino acid residues of citrine (candidate interactors^YN) and CUL1^NTD or ASK1 fused to the C-terminal 156–239 amino acid residues of citrine (^YCASK1 or ^YCCUL1^NTD). Co-expression of ^YCASK1 with DMR6^YN (DOWNY MILDEW RESISTANT 6), DLO1^YN (DMR6-LIKE OXYGENASE 1), GET3A^YN (GUIDED ENTRY OF TAIL-ANCHORED PROTEIN 3A), Oleosin^YN, OLE1^YN, CMPG2^YN (CYS, MET, PRO, AND GLY PROTEIN 2), MKK5^YN, MEKK1^YN (MAPK/ERK KINASE KINASE 1), RBR1^YN (RETINOBLASTOMA-RELATED PROTEIN 1), FBD^YN, or PhyA^YN, reconstituted citrine fluorescence, but not when ^YCASK1 co-expressed with GUS^YN control (Figs 2a, S4a). Co-expression of ^YCCUL1^NTD with PhyA^YN, Oleosin^YN, or GET3A^YN reconstituted citrine fluorescence, but not when ^YCCUL1^NTD co-expressed with GUS^YN control (Fig. S4a). Although EXO70A1 (AT5G03540.2) was identified as an interactor in our PL, we were unable to see reconstituted citrine fluorescence in the BiFC assay (Fig. S4a). These results indicated that 11 out of the 12 tested proteins interacted in the BiFC assay in planta.

Next, we selected a subset of confirmed BiFC interactors as potential novel SCF substrates for the Co-IP assay. To that end, 3xMyc epitope tag fused DMR6, DLO1, GET3A, Oleosin, and OLE1 were each co-expressed in N. benthamiana with 3xHA tag fused to ASK1. Myc-conjugated beads were utilized for the pull-down, followed by immunoblotting with HA antibodies. Immunoblot analysis confirmed a successful pull-down of HA-ASK1 by MYC-tagged targets from the plant extract (Fig. 2b). Myc-tagged citrine served as a control and did not pull-down HA-tagged ASK1 (Fig. 2b). Additionally, in a reciprocal experiment, Myc-tagged GET3A and Oleosin were also pulled down by HA-tagged CUL1^NTD using HA-conjugated beads (Fig. S4b).

To assess whether the interacting candidates are indeed SCF E3 ubiquitin ligase target substrates that undergo ubiquitination and proteasomal degradation, HA-tagged ubiquitin was co-expressed with Myc-tagged DMR6, DLO1, OLE1, Oleosin, GET3A, and PhyA in N. benthamiana. Myc-conjugated beads were used to pull down, followed by immunoblot using anti-HA antibodies. Immunoblot analysis showed high molecular weight polyubiquitin conjugates for each of the tested substrates (Fig. 3a). As the fate of SCF's polyubiquitinated substrates is typically proteasomal degradation, we tested the effect of MG132, a proteasome inhibitor (Teicher & Tomaszewski, 2015), on the degradation of Myc-tagged substrates. Remarkably, protein levels for all tested substrates were significantly elevated in the presence of MG132 (Fig. 3b), further corroborating that the identified interacting proteins serve as bona fide SCF ubiquitin ligase substrates.

Here, we present the SCF interactome landscape in planta and unveil new targets of the ubiquitin system by leveraging an optimized, tightly controlled PL strategy combined with protein interactions, polyubiquitination, and proteasomal degradation. While it has been previously demonstrated that PhyA and OLE1 undergo proteasomal degradation (Clough & Vierstra, 1997; Traver & Bartel, 2023), they have not been specifically linked to SCF ubiquitin ligase complex. It is possible that targets of the ubiquitin system can be regulated by distinct E3 ligase families. For instance, while OLE1 is identified as an SCF target in our study, it has recently been found to undergo ubiquitination mediated by the RING-type E3 MIEL1 (Traver & Bartel, 2023). Hence, it is conceivable that the regulation of OLE1 turnover can be regulated by multiple E3s, potentially dependent on factors such as differential expression or other post-translational modifications, such as phosphorylation. PhyA degradation, a pivotal process in plant photoreceptor regulation, has been regulated by the ubiquitin-proteasome pathway, as evidenced by various recognition domains for ubiquitination (Clough & Vierstra, 1997). Our research proposes an additional targeting mechanism of PhyA involving an SCF-type E3 ligase, and future studies are awaited to further reveal the associated F-box protein. Strikingly, DMR6, DLO1, Oleosin, and GET3A have not been previously shown to undergo ubiquitin-mediated proteasomal degradation. DMR6 and DLO1 were characterized as 2-oxoglutarate (2OG) and Fe(II)-dependent oxygenase superfamily proteins that catalyze the hydroxylation of the defense hormone salicylic acid (SA; Zhang et al., 2017; Zeilmaker et al., 2015). E3-mediated ubiquitination of SA-mediated immunity coactivator NPR1 is critical for mediating SA perception and establishment of systemic acquired resistance (Skelly et al., 2019; Zavaliev et al., 2020). This suggests a link between SCF E3-mediated ubiquitination and SA inactivation, further solidifying the ubiquitin system as a major player in SA signaling. Gaining further mechanistic insight into the ubiquitin-mediated degradation of DMR6 and DLO1 would have promising implications for crop improvement purposes, as loss-of-function for DMR6 and DLO1 has been shown to increase SA levels and resistance to pathogens (Zeilmaker et al., 2015; Zhang et al., 2017; Thomazella et al., 2021).

Notably, our proxitome profiled a relatively low number of F-box proteins, despite the highly efficient cis-biotinylation of ASK1, which serves as an intrinsic control for its expression and biotinylating activity. We can only speculate that the unexpectedly low abundance of F-box proteins might be due to limitations in the experiment's ability to capture some F-box proteins that assemble into SCF complexes during specific developmental stages, in specific tissues, or when plants are exposed to biotic and abiotic stimuli. It is also possible that effective biotinylation is influenced by the availability of exposed lysine residues to which biotin is transferred during PL. Another limitation of the proxitome could be related to our selection of ASK1 as a bait. While the rationale for selecting ASK1 is clearly justified by numerous reports of its prominent activity as an SCF adaptor, it cannot be ruled out that other underexplored ASK proteins, such as ASK2 and ASK11, may participate in the SCF complex and provide additional specificity for F-box selection in certain tissues, under specific environmental stimuli, and at different developmental stages (Gagne et al., 2002; Risseeuw et al., 2003).

Furthermore, it is possible that Arabidopsis ASK1 has dual, non-SCF functions. Recently, mammalian SKP1 has been shown to regulate a switch between autophagy and unconventional protein secretion, suggesting that ASK1 may have biological functions beyond its role as an adaptor protein within the SCF complex (Li et al., 2023). Therefore, the F-box proteins and other proteins we captured under our experimental conditions with ASK1 could represent a snapshot at a specific developmental stage and/or could be part of noncanonical SCF complexes. Further studies utilizing our transgenic lines described here will be valuable for capturing F-box proteins that assemble into SCF complexes and other targets that associate with ASK1 in an SCF-independent complex under various biological conditions.

In summary, our study has revealed numerous novel interactors and substrates of SCF-type ubiquitin ligase, underscoring the pivotal role of these enzymes in upholding homeostasis for facilitating optimal plant growth and development. Our research not only illuminates the SCF interactome landscape but also emphasizes the necessity for further elucidation of the functions of these interacting proteins within relevant biological contexts. Establishing a robust platform for capturing the E3 interactome has the potential to enhance our comprehension of protein turnover regulation through ubiquitination and degradation in response to various biotic and abiotic stressors.

None declared.

NS and SPD-K designed the research. FS and NH performed most of the experiments. YL assisted in cloning, optimization of TurboID assay, and transient expression experiments. NDM performed confocal microscopy. CM performed LC-MS/MS and analysis. CM and JWW performed proteomics data analysis and generated interaction networks. FS and NH wrote the original draft of the paper. NS, SPD-K, JWW and CM wrote, reviewed, and edited the paper.