ECLIPSE mediates selective degradation of inner nuclear membrane protein in plants

Abstract

The inner nuclear membrane (INM) hosts a unique set of membrane proteins essential for nuclear functions. Proteolytic removal of mislocalized or defective membrane proteins is of critical importance for maintaining the homeostasis and integrity of the INM. Previous studies revealed that INM protein degradation depends on a specialized ubiquitin–proteasome system termed INM-associated degradation (INMAD) in plants, requiring the CDC48 complex and the 26S proteasome for membrane protein retrotranslocation and destruction, respectively. However, details of the adaptor proteins that link membrane substrates to the CDC48/proteasome degradation machinery are still lacking in the pathway. Here, we report the discovery of ECLIPSE, a previously uncharacterized protein that may serve as such a molecular bridge in the degradation of the conserved INM protein SUN1. We demonstrate that ECLIPSE physically associates with CDC48 and exhibits strong transcriptional co-regulation with multiple established plant INMAD components. Mechanistically, ECLIPSE may act as an adaptor through its dual-domain architecture: its C-terminal PUB domain mediates direct interaction with CDC48, while its N-terminal ubiquitin-associated domain recognizes ubiquitinated INM substrates. Genetic and biochemical analyses further established that ECLIPSE is required for SUN1 protein degradation in Arabidopsis, supporting its role in the turnover of at least some INM proteins in plants.

ECLIPSE is a potential adaptor protein that links ubiquitinated inner nuclear membrane proteins to the CDC48 complex, facilitating their degradation and maintaining nuclear membrane integrity in plants.

Introduction

The nuclear envelope (NE) is a double-layered membrane structure comprising the inner nuclear membrane (INM) and the outer nuclear membrane (ONM), with the latter being continuous to and sharing certain components with the endoplasmic reticulum (ER). In contrast to the ONM, the INM is characterized by a distinct population of transmembrane proteins, which compositionally and functionally distinguish the INM from other endomembranes. While the INM proteome in plants remains to be fully elucidated, several of its constituents have been shown to play critical roles in essential nuclear functions, including chromatin organization, nuclear signaling, and transcriptional regulation (Ané et al., 2004; Riely et al., 2007; Capoen et al., 2011; Oda and Fukuda, 2011; Murphy and Bass, 2012; Wong et al., 2014; Varas et al., 2015; Charpentier et al., 2016; Meier et al., 2017; Kang et al., 2022; Tang et al., 2022).

The SAD1 AND UNC84 (SUN) proteins are among the first reported and most well-studied INM proteins in plants (Van Damme et al., 2004; Graumann and Evans, 2010; Graumann et al., 2010; Murphy et al., 2010; Graumann, 2014). SUN1/2 proteins exhibit a special structural arrangement, with their N-termini associating with the nucleoskeleton within the nucleoplasm, while their C-termini bind to ONM integral proteins known as Klarsicht/ANC-1/Syne Homology (KASH) proteins in the perinuclear region. KASH proteins, in turn, associate with the cytoskeleton, and these intricate connections form the linker of nucleoskeleton and cytoskeleton (LINC) complex, bridging between the nucleus and the cytoplasm, and facilitating mechanical force transmission and cellular signaling across the NE (Sosa et al., 2012; Zhou et al., 2014). SUN proteins have been implicated in various cellular processes in plant cells, including the regulation of nuclear morphology, nuclear movement, nuclear positioning in pollen tubes, and chromosome dynamics during the early stages of meiosis I (Zhou et al., 2012, 2015a, b, c; Liang et al., 2015; Varas et al., 2015; Moser et al., 2020). In addition to SUN proteins, PLANT NUCLEAR ENVELOPE TRANSMEMBRANE 2 (PNET2), a six-pass transmembrane INM protein, was recently identified as an essential component of the nuclear lamina in plants. PNET2 was shown to play a crucial role in chromatin tethering at the nuclear periphery, a function that is essential for regulating gene expression that coordinates growth and stress responses in Arabidopsis (Tang et al., 2022).

Emerging research highlights that INM proteins are subjected to selective removal and active degradation, a process critical for their functional homeostasis. Transmembrane proteins, particularly those that are isolated within certain endomembrane organelles, such as the ER and chloroplasts, are excluded from the vacuolar degradation pathway. As a result, they are degraded through the conserved ubiquitin–proteasome system (UPS) (Avci and Lemberg, 2015; Sun and Jarvis, 2023). Accumulating evidence suggests that eukaryotes have developed sophisticated, UPS-mediated, membrane-associated degradation systems for removing integral proteins at various organellar membranes. Among them, the ER-associated protein degradation (ERAD) pathway is the most extensively characterized (Lopata et al., 2020). More recently in plants, additional pathways have been elucidated, including the chloroplast-associated protein degradation (CHLORAD) pathway, responsible for chloroplast membrane protein degradation (Ling et al., 2012, 2019, 2021; Ling and Jarvis, 2015; Shanmugabalaji and Kessler, 2019), and the lipid droplet-associated degradation (LDAD) pathway, which removes membrane proteins from lipid droplets (Deruyffelaere et al., 2018; Kretzschmar et al., 2018).

These membrane-associated degradation pathways share common features. First, they rely on organelle-specific E3 ubiquitin ligases, which are either integral membrane proteins or membrane associated, to ubiquitinate target membrane proteins for degradation. Following substrate ubiquitination, a conserved retrotranslocation mechanism comes into play, involving the AAA ATPase CELL DIVISION CYCLE 48 (CDC48) protein and its cofactors UBIQUITIN FUSION DEGRADATION 1 (UFD1) and NUCLEAR PROTEIN LOCALIZATION 4 (NPL4). This complex recognizes ubiquitinated membrane substrates through adaptor proteins that link the substrates to CDC48 (Barthelme and Sauer, 2016; van den Boom and Meyer, 2018; Lopata et al., 2020). The CDC48–UFD1–NPL4 complex then mediates the removal of ubiquitinated substrates from the membrane through retrotranslocation, using ATP hydrolysis to overcome the energetic barrier imposed by hydrophobic transmembrane domains and effectively extracting substrate proteins from the lipid layer. Subsequently, the extracted polypeptides are degraded by the 26S proteasome (Berner et al., 2018; Strasser, 2018).

Despite extensive knowledge of membrane protein degradation in various organelles via the UPS, little is known about this process at the nucleus. However, disruptions to nuclear morphology and various genetic diseases have been tightly linked to the excessive accumulation of nuclear membrane proteins, particularly the INM proteins, underscoring the functional importance of INM protein turnover (Burke and Stewart, 2014; Jevtić et al., 2014). Studies in yeasts have revealed the existence of a specialized UPS termed the inner nuclear membrane-associated degradation (INMAD) pathway, which is responsible for INM protein degradation. In yeast INMAD, amino acid sensor-independent 1 (Asi1) and its homolog Asi3 are multi-pass transmembrane E3 ubiquitin ligases that localize to the INM and possess a Really Interesting New Gene (RING) domain, which catalyzes substrate ubiquitination (Foresti et al., 2014; Khmelinskii et al., 2014). Asi1 and Asi3 form a complex with Asi2, which assists with substrate recognition (Foresti et al., 2014; Natarajan et al., 2020). The yeast INMAD has been shown to maintain the identity of the INM proteome, by eliminating orphaned transmembrane protein subunits that mislocalized to the INM, as well as misfolded or damaged INM proteins (Smoyer and Jaspersen, 2019; Smoyer et al., 2019; Natarajan et al., 2020). However, Asi genes are not found outside of the fungi kingdom, leaving the mechanisms of INM protein degradation in higher eukaryotes, including plants, completely unknown.

Previously, research from our group extended the INMAD system to plants. We demonstrated that SUN1, a conserved INM protein, undergoes proteasome-dependent degradation within the nucleus in Arabidopsis (Huang et al., 2020). Furthermore, proximity labeling proteomics using SUN1 as bait identified the CDC48–UFD1–NPL4 complex alongside three closely related UBIQUITIN REGULATORY X (UBX) proteins in the PLANT UBX DOMAIN-CONTAINING (PUX) family, including PUX3, PUX4, and PUX5. These three PUX homologs were shown to bind directly with CDC48. However, in contrast to most other UBX proteins, which were reported to function as adaptors between substrates and CDC48, PUX3/4/5 appear to function redundantly in protecting SUN1 from degradation (Huang et al., 2020; Zhang et al., 2021). Therefore, the adaptor protein(s) that bridges ubiquitinated membrane substrates with CDC48, an essential INMAD component, has yet to be identified.

In this study, we identify ECLIPSE, a previously uncharacterized Arabidopsis protein, as a candidate adaptor in the plant INMAD pathway for SUN1 degradation. We demonstrate that ECLIPSE physically binds to CDC48 through a PNGase/UBA OR UBX (PUB) domain located at its C-terminus. We also show that the N-terminal ubiquitin-associated (UBA) domain of ECLIPSE is required for its interaction with SUN1 at the INM. These interactions position ECLIPSE as a potential molecular bridge between ubiquitinated substrates and the CDC48 complex. Moreover, we present evidence that the function of ECLIPSE is indispensable for the proteasomal degradation of SUN1. These findings collectively suggest a possible molecular mechanism by which ECLIPSE facilitates the recognition and processing of ubiquitinated substrates in the plant INMAD pathway.

Materials and methods

Plant material and growth conditions

The ecotype of all Arabidopsis plants used in this study is Col-0. The eclipse T-DNA line (SAIL_91_H06) was obtained from the Arabidopsis Biological Resource Center (ABRC). The HA-BioID2-SUN1 transgenic line was previously published (Huang et al., 2020). The isogenic HA-BioID2-SUN1 lines in the wild-type (WT) and eclipse mutant background were obtained through genetic cross and confirmed via PCR-based genotyping. The 35S:ECLIPSE-eGFP lines are in the WT background and were generated by floral dipping using Agrobacterium tumefaciens GV3101 and selected on half-strength Murashige and Skoog (1/2 MS) agar (0.8%) plates containing hygromycin B (25 μg ml^–1). All seeds used were surface-sterilized with 70% ethanol for 5 min followed by 10% (v/v) bleach for 10 min and five rinses with sterile water. Seeds were germinated on MS plates after stratifying at 4 °C for 48 h and grown under a 16 h light/8 h dark–light cycle at 22 °C with light intensity 90 µmol m^–2 s^–1.

Plasmid construction

For yeast two-hybrid (Y2H) assays, the full-length cDNA sequences of ECLIPSE (AT1G04850), PUX5, as well as a truncation of ECLIPSE lacking the PUB domain (last 107 amino acids; truncation amino acids 1–306) were cloned into the pGBKT7 vector containing the binding domain (BD) via the In-Fusion cloning system (ClonExpress II One Step Cloning Kit, Vazyme). Similarly, the full-length CDC48A cDNA, the N-terminus of CDC48A (first 191 amino acids), the D1 domain (amino acids 211–461), and the D2-lid domain (amino acids 484–809) were cloned into the pGADT7 vector containing the activation domain (AD).

To generate green fluorescent protein (GFP)-tagged constructs, the full-length genomic DNA sequences of ECLIPSE and PUX5 were cloned into a modified pCambia1300 binary vector with an eGFP tag at the C-terminus using In-Fusion cloning. The GFP-tagged ECLIPSE construct was transformed into Agrobacterium and used either for transient protein expression or to generate the transgenic Arabidopsis lines. To generate HA-tagged constructs for transient assays, the full-length genomic DNA sequence of SUN1 was cloned into a modified pEarleyGate binary vector with an HA tag at the N-terminus and a TurboID tag at the C-terminus through In-Fusion cloning. For bifluorescence complementarion (BiFC) assays, the full-length genomic DNA sequences of SUN1, ECLIPSE, and IMB2, together with a truncated sequence of ECLIPSE lacking the UBA domain (first 151 amino acids), were cloned into a modified pCambia1300 binary vector carrying the N- or C-terminal fragments of yellow fluorescent protein (nYFP or cYFP, respectively). SUN1 was tagged with nYFP at the N-terminus whereas ECLIPSE, ECLIPSEΔUBA, and IMB2 were tagged with cYFP at the C-terminus. All constructs were used for transient expression in Nicotiana benthamiana via Agrobacterium-mediated infiltration.

All primers used for cloning are provided in Supplementary Table S1.

Phylogenetic analysis

The full-length amino acid sequence of ECLIPSE and CDC48A were used to run a protein BLAST search (https://blast.ncbi.nlm.nih.gov/Blast.cgi) with the following species: Chlamydomonas reinhardtii, Physcomitrium patens, Ceratopteris richardii, Ananas comosus, Zea mays, Oryza sativa, Triticum aestivum, Populus trichocarpa, Solanum lycopersicum, Glycine max, Drosophila melanogaster (not for CDC48A), and Homo sapiens. The top scoring homologs from each species were selected and aligned with AtECLIPSE employing Multiple Alignment using Fast Fourier Transform (MAFFT) (https://mafft.cbrc.jp/alignment/software/). Alignment visualization and curation were done using JalView (https://www.jalview.org/). Tree inference was performed through Fasttree on Linux using the approximately-maximum-likelihood method and the Shimodaira–Hasegawa method for branch support (https://arkinlab.bio/fasttree/). Tree rendering and manipulation were performed using FigTree (http://tree.bio.ed.ac.uk/software/figtree/). Sequence logos for the PUB and CDC48 C-terminal motifs were generated using WebLogo3 (https://weblogo.threeplusone.com/) to show residue frequency based on multiple sequence alignment.

Yeast two-hybrid assay

Y2H assays were performed using the Matchmaker Two-Hybrid System 2 (Clontech) and according to the manufacturer’s instructions. The BD and AD constructs were transformed into the AH109 and Y187 yeast strains, respectively. Transformed AH109 and Y187 colonies expressing the corresponding proteins were mated in 2× YPDA medium at 30 °C for 24 h and plated for selection on double (SD/-Leucine/-Tryptophan), triple (SD/-Leucine/-Tryptophan/-Histidine), and quadruple (SD/-Leucine/-Tryptophan/-Histidine/-Adenine) synthetic dropout media at 30 °C for 5–7 d before images were taken.

Co-immunoprecipitation

Co-immunoprecipitation (co-IP) assays were carried out using transiently expressed protein (Gu et al., 2016). Agrobacterium carrying relevant constructs were co-infiltrated on the abaxial side of leaves of 3- to 5-week-old N. benthamiana. Two days post-infection, leaves were collected, frozen in liquid nitrogen, and ground to a fine powder with a mortar and pestle. Tissue powder was weighed, resuspended in protein extraction buffer [50 mM Tris–HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 0.4% Triton X-100, 1 mM DTT, 1× Protease Inhibitor, 20 mM N-ethylmaleimide], and incubated on ice for 15 min before centrifugation at 18 000 g for 15 min at 4 °C. For co-IP, 1 ml of supernatant was incubated for 2 h with 15 µl of GFP-Trap Agarose beads at room temperature (Chromotek). Beads were then pelleted at 1500 g for 5 min at 4 °C and washed five times with wash buffer [50 mM Tris–HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 0.1% Triton X-100, 1 mM DTT, 20 mM N-ethylmaleimide] before being resuspended in 50 µl of loading buffer (2× SDS with 10% β-mercaptoethanol). Boiled samples were separated via SDS–PAGE and subjected to immunoblotting using anti-GFP antibody (Takara Bio, Cat #632381, dilution 1:3000) and anti-HA antibody (Roche, Cat #11867431001, dilution 1:3000).

Fluorescence imaging

ECLIPSE subcellular localization was determined using Agrobacterium-mediated transient expression in N. benthamiana leaves or using 1-week-old transgenic Arabidopsis plants. Infected N. benthamiana epidermal leaves were used for confocal microscopy with a Zeiss LSM710 inverted confocal laser scanning microscope with ZEN software. eGFP fluorescence was excited by the 488 nm argon laser and detected using a custom 522–545 nm band-pass emission filter, whereas mCherry fluorescence was excited by the 594 nm laser and detected using a custom 595–620 nm band-pass emission filter.

For BiFC experiments, Agrobacterium carrying BiFC constructs were co-infiltrated into N. benthamiana leaves, which were then used for confocal microscopy. Confocal laser scanning microscopy was performed using a Zeiss LSM880 inverted confocal microscope with ZEN software. sYFP fluorescence was excited with a 514 nm laser and detected using a custom 525–540 nm band-pass emission filter.

Quantitative reverse transcription–PCR

TRIzol reagent (Invitrogen, Cat #15596026) was used to extract RNA from 1-month-old Arabidopsis leaves. cDNA was synthesized using the Maxima First Strand cDNA Synthesis kit with the dsDNase Kit (Thermo Scientific, Cat #K1672). The quantitative PCR (qPCR) was run on a CFX96TM Real-Time PCR Detection System (Bio-Rad) using the SYBR Green PCR Master Mix (Thermo Fisher, Cat # 4309155). Results were confirmed by using two reference genes, namely Actin2 and UBQ5. All primers used for qPCR are provided in Supplementary Table S1.

In vivo protein degradation assay

Ten-day-old Arabidopsis seedlings were treated with 150 µM cycloheximide (CHX) for the indicated lengths of time. Approximately 20 seedlings of similar sizes were included in one sample. Sampled seedlings were blotted dry and weighed to standardize biomass across samples. Seedlings were then flash-frozen, ground to a fine powder, and resuspended in protein extraction buffer as described above. After thorough vortexing, samples were centrifuged at 16 200 g for 15 min at 4 °C. The supernatant was collected and mixed with 2× SDS buffer with 10% β-mercaptoethanol. Samples were then boiled and separated by SDS–PAGE followed by immunoblotting using anti-HA antibody and anti-actin antibody.

Bioinformatic analysis

Proteomic data of ECLIPSE were obtained from previously published proximity labeling proteomic datasets using SUN1, WIP1, and nucleoporins as baits, and are available through the Proteomics Identification Database (PRIDE) (identifiers PXD015919 and PXD015920) (Huang et al., 2020; Tang et al., 2020). For ECLIPSE co-expression analysis, candidate genes selected from the overlapping SUN1 and PUX5 proxiomes were used. Alternatively, UPS-associated genes and a pool of random genes generated through University of Toronto’s Bio-Array Resource (BAR) Random AGI ID List Generator Tool (https://bar.utoronto.ca/) were used. The Pearson’s correlation coefficients based on microarray data were retrieved from CORNET (https://bioinformatics.psb.ugent.be/cornet/versions/cornet3.0/). Protein domain prediction was obtained from SMART (http://smart.embl-heidelberg.de/). Data were visualized using Python’s Seaborn package. All Linux commands were executed on the Savio high performance computing cluster of the University of California at Berkeley.

The structure of the ECLIPSE PUB (amino acids 307–413) domain was predicted using AlphaFold2 through ColabFold v1.5.5 (Mirdita et al., 2022) and downloaded as a PDB format file. The structure of the human UBXD1 domain was submitted as a template (PDB: 6SAP) (Blueggel et al., 2019). Structural overlay and visualization were performed on Chimera X-1.8 on Windows through the Matchmaker tool and using human UBXD1-PUB as the reference structure.

Results

ECLIPSE is a component of the plant inner nuclear membrane-associated degradation system

To elucidate additional critical components of the plant INMAD pathway, we conducted a reanalysis of previously obtained proximity labeling proteomic data using SUN1 (an INMAD substrate) and PUX5 (an INMAD regulator) as bait proteins. We focused on identifying proteins present in both SUN1 and PUX5 proxiomes, as these shared interactors have a higher probability of being an integral component to the INMAD system (Fig. 1A). With a combined analysis of the two MS datasets applying cut-offs (fold change >2, P-value <0.05) compared with mock-treated samples for each bait, we identified 31 candidates as shared proximal proteins for SUN1 and PUX5 (Fig. 1B). Among them were previously identified plant INMAD components, including UFD1B and UFD1C, part of the CDC48–UFD1–NPL4 complex, as well as PUX4 and PUX5.

Fig. 1.

ECLIPSE is closely associated with plant INMAD components. (A) Schematic diagram illustrating the identification of potential INMAD components using proximity labeling with SUN1 and PUX5 as baits. SUN1, a substrate of the INMAD pathway, undergoes polyubiquitination by unidentified E3 ligases. The ubiquitinated SUN1 is recognized by a potential adaptor protein, which facilitates the recruitment of CDC48 for subsequent retrotranslocation and degradation of SUN1. PUX5 is a previously reported negative regulator of the INMAD pathway, which interacts with CDC48 and functions to prevent SUN1 degradation. (B) Reanalysis of previously published MS data obtained from proximity labeling proteomics using HA-BioID2-SUN1 and PUX5-BioID2-HA transgenic plants. Specific proteins probed by each bait were identified using biotin mock-treated transgenic plants as controls. Peptide intensity data from three biological replicates for each sample were normalized and subjected to ratiometric analysis. Preys with a fold change >2 (red dashed lines) and an adjusted P-value <0.05 were considered specific preys (blue dots). The 31 overlapping candidates identified by both baits are labeled in red dots and shown in the Venn diagram. MS data are available through PRIDE (PXD015919 and PXD015920). (C) Illustration of the ECLIPSE protein domain architecture. Protein domains were predicted by SMART (http://smart.embl-heidelberg.de/).

Of particular interest among the candidates was a previously uncharacterized protein (AT1G04850) that contains a UBA domain close to its N-terminus and a PUB domain at its C-terminus (Fig. 1C). The UBA domain is known for its capacity to bind both mono- and polyubiquitin, and is common to PUX proteins (Meyer et al., 2012; Zhang et al., 2021). On the other hand, the PUB domain has been reported to directly interact with CDC48 in humans. For example, the N-terminal PUB domain of UBIQUITIN REGULATORY X DOMAIN PROTEIN 1 (UBXD1) interacts with p97, the human ortholog of CDC48, and is involved in the recruitment of the proteasome following the exit of substrates from p97 (Blueggel et al., 2019, 2023). This distinctive domain architecture makes the protein a promising candidate as an adaptor, with the potential to facilitate the recognition of ubiquitinated SUN1 while engaging CDC48 for subsequent substrate retrotranslocation and proteosome degradation. We thus named it ECLIPSE.

ECLIPSE is co-expressed with plant inner nuclear membrane-associated degradation components and localizes within the nucleus

In addition to the protein domain architecture, another compelling feature that distinguishes ECLIPSE from other candidates is its robust co-expression with known INMAD components. For example, among the 31 shared proximal proteins identified in SUN1 and PUX5 proxiomes, ECLIPSE is clustered in a co-expression group with PUX5, UFD1B, and UFD1C (Fig. 2A; Supplementary Table S2). This cluster also contains plant nuclear lamina constituents, such as CRWN genes, which we previously showed to be involved in the INMAD process (Huang et al., 2020). To further strengthen the co-expression between ECLIPSE and components of the INMAD pathway, we examined the co-expression patterns of ECLIPSE with known plant genes that function in the UPS, along with a set of randomly selected genes as control. Remarkably, the resulting clustered heatmap revealed that except for PUX4, ECLIPSE is strongly and specifically co-expressed with all the previously identified plant INMAD players, including all three components of the CDC48–UFD1–NPL4 complex and PUX3/5 (Fig. 2B; Supplementary Table S2). This co-expression pattern was not observed with randomly selected genes, suggesting that it is specific. This co-expression profile further underscores the potential involvement of ECLIPSE in the INMAD pathway.

Fig. 2.

ECLIPSE displays robust co-expression with INMAD genes and localizes in the nucleus. (A) Gene co-expression pattern of the 31 overlapping proteins identified in both SUN1 and PUX5 proxiomes. Pearson’s correlation coefficients, calculated using gene expression data from the CORNET database (version 3.0 for Arabidopsis), are retrieved from the CORNET co-expression tool and presented in a clustered heatmap. (B) Gene co-expression patterns of known plant INMAD-related genes along with 14 randomly genes generated using the Random AGI ID List Generator Tool (BAR). Correlation coefficient data are provided in Supplementary Table S2. (C) Transient coexpression of ECLIPSE–eGFP and free mCherry in Nicotiana benthamiana. Leaf epidermal cells were imaged. Scale bar=100 μm. (D) Fluorescence imaging of ECLIPSE–eGFP subcellular localization in Arabidopsis root cells. The left panel shows root tips of 1-week-old 35S:ECLIPSE-eGFP transgenic plants. Scale bar=50 μm. A magnified view of nuclei is presented in the right panel. Scale bar=5 μm. (E) Phylogenetic analysis of ECLIPSE and its homologs across various animal and plant species using protein sequences. Homologs with the highest sequence similarity with ECLIPSE in each species were selected for plotting. The phylogenetic tree was inferred using FastTree with the approximately maximum-likelihood method, and branch support was assessed using the Shimodaira–Hasegawa-like test and displayed as node labels. Representative protein domain architectures corresponding to each subclade are displayed on the right.

To corroborate the involvement of ECLIPSE in INMAD, we also characterized its subcellular localization. We transiently expressed a C-terminal fusion of ECLIPSE with eGFP driven by the 35S cauliflower mosaic virus promoter (35S:ECLIPSE-eGFP) in N. benthamiana. Fluorescence microscopy analysis of transformed epidermal cells revealed that ECLIPSE–eGFP exhibits a nucleocytoplasmic distribution, with a predominant signal observed within the nucleus (Fig. 2C). We subsequently generated Arabidopsis stable transgenic lines expressing the same construct. Similarly, in independent transgenic seedlings, ECLIPSE–eGFP is primarily localized to the nucleus (Fig. 2D). In some cells, ECLIPSE forms visible nuclear speckles that were stochastically distributed across the nucleus. These results indicate that ECLIPSE is likely to function mainly in the nucleus, in line with its potential role in INMAD and supporting its detection in proximity labeling proteomics by SUN1 and PUX5.

Phylogenetic analysis revealed that homologs of ECLIPSE are widely distributed across plant taxa, with homologous sequences identified across nine species spanning most groups of land plants as well as an algal species (Fig. 2E). Additionally, sequence homology was observed in Drosophila and human UBX domain-containing proteins, with HsUBXN6 showing 45% sequence similarity (e-value=6×10^–15). The phylogenetic tree exhibits a well-supported grouping among monocots and dicots, with ferns, bryophytes, and algae being progressively distant. Interestingly, the predicted domain architecture of ECLIPSE homologs shows remarkable conservation within land plant species, characterized by an arrangement of two N-terminal zinc-finger domains, an UBA domain, an extended coiled-coil domain, and a C-terminal PUB domain (Fig. 2E). This conserved multidomain organization suggests coordinated functionality among these domains. However, the algal and animal homologs lack both zinc-finger and UBA domains at the N-terminus, suggesting that the ubiquitin binding capacity of ECLIPSE evolved uniquely in land plants. The presence of ECLIPSE and its homologs across diverse plant lineages underscores their potential functional significance in plant cells.

ECLIPSE physically interacts with the C-terminus of CDC48 via its PUB domain

To investigate the role of ECLIPSE within the UPS system, especially as a potential adaptor protein between INM substrates and CDC48, we employed Y2H assays to examine the physical interaction between ECLIPSE and CDC48. As positive controls, we included PUX5, which we have previously shown to interact with CDC48 (Huang et al., 2020). Our results indicate that ECLIPSE, like PUX5, is able to directly interact with CDC48A, a principal paralog of CDC48 in Arabidopsis (Fig. 3A). Moreover, this interaction depends on the PUB domain of ECLIPSE (Fig. 3B). This finding aligns with previously reported interactions between the PUB domain of UBXD1 and p97 in humans, suggesting a conserved role for the PUB domain in mediating interactions with CDC48 across eukaryotic lineages.

Fig. 3.

ECLIPSE directly binds the C-terminus of CDC48 through its PUB domain. (A) Yeast two-hybrid (Y2H) analysis using ECLIPSE and PUX5 as the bait and CDC48A as the prey. (B) Y2H analysis using the full-length ECLIPSE and ECLIPSE^ΔPUB (amino acids 1–306) as the bait and CDC48A as the prey. (C) Y2H analysis using ECLIPSE and PUX5 as the bait and truncations of CDC48A consisting of either its N-terminal domain (amino acids 1–191), the D1 domain (amino acids 211–461), or the D2 domain (amino acids 484–809) as the prey. Yeasts were grown on double dropout (DDO) (SD/-Leu-Trp), triple dropout (TDO) (SD/-Leu-Trp-His), and quadruple dropout (QDO) (SD/-Leu-Trp-His-Ade) media. Empty vectors were used as negative controls. (D) Amino acid sequence consensus of the CDC48 C-terminal PIM motif (last 10 residues, lower panel) and the PIM-binding pocket of ECLIPSE (upper panel), derived from multiple sequence alignment of 11 plant and algal species used in the phylogenetic analysis in Fig. 2. Residues and their positions in Arabidopsis and human homologs are shown at the top and bottom of each panel, respectively. Sequence logos were generated using WebLogo3. (E) Structural overlay of the UBXD1 PUB domain (PDB #6SAP; colored in beige) and the predicted ECLIPSE PUB domain (amino acids 307–413) by AlphaFold2 (ColabFold v1.5.5). The ECLIPSE PUB structure is colored according to per-residue predicted local distance difference test (pLDDT) scores, with 0 (red), 50 (orange), 70 (yellow), 90 (light blue), and 100 (blue) indicating varying levels of prediction confidence.

Substrate unfolding by the CDC48 complex is regulated through its N-terminal domain, and most characterized CDC48 cofactors, including PUX proteins, bind to the N-terminal regulatory region of CDC48 through their UBX domains (Zhang et al., 2021). As ECLIPSE is among the few proteins containing a PUB rather than a UBX domain for interaction with CDC48, we further determined its CDC48-binding sites. While the UBX-containing PUX5 interacted with the N-terminus of CDC48A as expected, ECLIPSE showed no interactions with this domain (Fig. 3C). Instead, ECLIPSE displayed a strong interaction with the C-terminus of CDC48A, representing a distinct interaction pattern.

Previous structural studies have characterized the interaction between the PUB domain of human UBXD1 and the PUB-interacting motif (PIM) in p97 (Blueggel et al., 2019). The PIM motif is made up of the C-terminal 10 amino acids of p97 (TEDNDDDLYG) that are cradled inside a hydrophobic pocket of UBXD1–PUB. This interaction is mediated by electrostatic interactions between multiple charged amino acids in both PUB and PIM, as well as the π-stacking of UBXD1–Y181/Y194 with p97–Y805 (Blueggel et al., 2019). Analysis the ECLIPSE PUB motif and CDC48 PIM motif across 11 plant and algal species demonstrated high sequence conservation, especially charged residues and tyrosine residues for π-stacking (Fig. 3D). Structural prediction also revealed an almost complete overlay of the PUB domain from AtECLIPSE and HsUBXD1 (Fig. 3E). These findings reinforce the binding of ECLIPSE to the C-terminus of CDC48 and indicate an evolutionarily conserved interaction mechanism.

ECLIPSE associates with a conserved plant inner nuclear membrane-associated degradation substrate SUN1

The functional specificity of CDC48 across different subcellular compartments is primarily determined by its adaptors. For instance, PUX10 localizes to lipid droplets and chloroplasts, where it recruits CDC48 to facilitate membrane protein degradation in these two compartments (Deruyffelaere et al., 2018; Kretzschmar et al., 2018; Li and Jarvis, 2024). In contrast, PUX7, PUX8, PUX9, and PUX13 associate with ATG8, recruiting CDC48 to autophagosomes for destruction (Marshall et al., 2019). By mining published proxiomes from various nuclear membrane proteins, we observed a preferential association of ECLIPSE with INM baits. Specifically, INM transmembrane proteins SUN1 and NEAP1, when used as baits, consistently detected significantly higher levels of ECLIPSE compared with the ONM protein WIP1 and the nuclear pore complex (NPC) component Nup93a (Fig. 4A).

Fig. 4.

ECLIPSE binds SUN1 at the INM. (A) Reanalysis of MS data from previously published proximity labeling proteomics using different baits targeting the inner nuclear membrane (INM), outer nuclear membrane (ONM), and nuclear pore complex (NPC) (PXD015919). Biotin-treated WT non-transgenic plants (NT) served as controls. Normalized peptide–spectrum match (PSM) values for INMAD components are plotted. Statistical analysis of protein enrichment was performed using mock-treated transgenic lines as controls. (B) Co-immunoprecipitation assay using N. benthamiana leaves transiently co-expressing free GFP or ECLIPSE–eGFP together with HA-SUN1-TurboID. Protein extract was immunoprecipitated with GFP-Trap agarose beads before immunoblotting with anti-HA and anti-GFP antibodies. (C) Bimolecular fluorescence complementation (BiFC) assay examining interactions between ECLIPSE and SUN1 or IMB2, as well as between truncated ECLIPSE^ΔUBA (amino acids 152–413) and SUN1. BiFC constructs were transiently co-expressed in N. benthamiana leaves with free mCherry to label the nuclei. Scale bars=60 μm (top and center) and 50 μm (bottom).

Nevertheless, we were unable to detect physical interactions between SUN1 and ECLIPSE using Y2H assay. Considering that the interaction between ECLIPSE and SUN1 may be transient, require specific modifications (e.g. ubiquitination), or necessitate additional cofactors not present in the yeast system, we transiently co-expressed SUN1 with ECLIPSE in N. benthamiana and performed co-IP assays. We detected association between ECLIPSE and SUN1 (Fig. 4B). In contrast, SUN1 was not co-immunoprecipitated with free GFP.

To confirm this interaction in vivo, we further conducted a BiFC assay between ECLIPSE and SUN1 using transient expression. BiFC signal was detected at both the nuclear membrane and periphery ER, probably due to overexpression. Nevertheless, BiFC signal was not observed when IMPORTIN-β2 (IMB2), a nucleocytoplasmic distributed karyopherin protein, was co-expressed with ECLIPSE (Fig. 4C). Moreover, deletion of the UBA domain of ECLIPSE effectively disrupted its interaction with SUN1 on the nuclear membrane, implying that ECLIPSE binds ubiquitinated SUN1 and consistent with the proposed role of the UBA domain in substrate binding. Taken together, these data show that ECLIPSE associates with SUN1 at the INM and its UBA domain is necessary for this interaction.

ECLIPSE is required for ubiquitin–proteasome system-dependent SUN1 degradation

We previously demonstrated that SUN1 undergoes constitutive degradation via INMAD using 35S:HA-BioID2-SUN1 transgenic Arabidopsis plants (Huang et al., 2020). To elucidate the functional importance of ECLIPSE in SUN1 degradation, we obtained a T-DNA insertional mutant line for ECLIPSE (SAIL_91_H06) from the ABRC (Fig. 5A). Thermal Asymmetric Interlaced (TAIL) PCR followed by sequencing of the PCR product verified a single T-DNA insertion in the genome at the ECLIPSE locus, and quantitative reverse transcription–PCR (RT–qPCR) analysis confirmed a marked reduction in ECLIPSE expression in this mutant line, which we designated as eclipse-1 (Fig. 5B) (Liu et al., 1995). Moreover, the location of the T-DNA insertion indicates that the eclipse-1 mutant is unlikely to produce functional gene products. We then crossed eclipse-1 with a single-copy 35S:HA-BioID2-SUN1 transgenic line, yielding isogenic lines in the WT and eclipse-1 mutant background. Immunoblot analysis measuring the SUN1 protein level was performed using non-segregating T₄ plants. We observed a significantly higher steady-state accumulation of SUN1 in eclipse-1 seedlings compared with WT controls (Fig. 5C), suggesting that ECLIPSE indeed plays a role in regulating SUN1 protein homeostasis.

Fig. 5.

ECLIPSE is involved in SUN1 degradation. (A) Gene structure of AtECLIPSE and the T-DNA insertion site of the eclipse-1 mutant line. Primers used for qPCR analysis are indicated by arrows. (B) Gene expression of ECLIPSE relative to Actin in the eclipse-1 mutant, normalized to that of the WT. One-month-old leaves were used for mRNA preparation. RT–qPCR data are presented as the mean ±SD (n=3 biological replicates), with statistical analysis performed using a two-tailed Student’s t-test. Similar results were obtained using UBQ5 as the reference gene. (C) The steady-state accumulation of SUN1 protein measured in isogenic 35S:HA-BioID2-SUN1 transgenic plants in WT or eclipse-1 mutant backgrounds using immunoblots. The protein level of SUN1 relative to actin was normalized to 1 in the WT background (n=3 biological replicates). (D) In vivo protein chase experiment measuring the SUN1 protein degradation rate using isogenic 35S:HA-BioID2-SUN1 transgenic plants in WT or eclipse-1 mutant backgrounds. Plants were treated with 150 μM CHX, and protein was extracted at the indicated times post-CHX treatment. Immunoblotting was performed using anti-HA and anti-actin antibodies. (E) Time-course densitometric analysis of SUN1 levels, presented as bar graphs with the mean ±SD (n=3 biological replicates, two-tailed Student’s t-tests). The expression level of SUN1 relative to actin was normalized to 1 at the 0 time point in each background.

To further investigate the role of ECLIPSE in SUN1 turnover, we performed chase experiments by inhibiting de novo SUN1 protein synthesis with CHX. 35S:HA-BioID2-SUN1 seedlings were sampled at multiple time points after being treated with CHX, and SUN1 protein levels were assessed via immunoblotting. In independent experiments, we constantly observed that the rate of SUN1 protein degradation was significantly slower in the eclipse-1 mutant compared with the WT (Fig. 5D, E). These data demonstrate that SUN1 is stabilized in the absence of ECLIPSE, supporting a crucial role for ECLIPSE in mediating SUN1 degradation, as a potential substrate adaptor protein.

Discussion

Eukaryotes have evolved organelle-specific branches of membrane-associated protein degradation systems to remove transmembrane proteins in various organelles. Besides the conserved and well-characterized ERAD pathway, the CHLORAD system, the LDAD system, and the INMAD system were recently discovered in plants (Deruyffelaere et al., 2018; Kretzschmar et al., 2018; Ling et al., 2019, 2021; Huang et al., 2020; Calvanese and Gu, 2022). A common feature of these pathways is their dependence on CDC48, a key protein responsible for extracting proteins from membranes and directing them to the 26S proteasome for degradation. PUX proteins have emerged as essential adaptors that guide CDC48 to specific organelle membranes and regulate substrate selection. In Arabidopsis, several PUX proteins function as cofactors, adapting organelle-specific substrates and interacting with the N-terminal domain of CDC48 through UBX or UBX-like domains, thereby facilitating precise membrane protein degradation across distinct cellular compartments (Deruyffelaere et al., 2018; Kretzschmar et al., 2018; Marshall et al., 2019).

Studies here identify a previously unknown CDC48 cofactor in plants that utilizes a PUB domain rather than the UBX domain for CDC48 interaction, specifically in the context of INMAD. In humans, the PUB domain was first characterized as a p97-binding module for peptide N-glycanase (PNGase) (Allen et al., 2006). The PNGase PUB domain was shown to bind the C-terminal PIM motif of p97, facilitating the turnover of misfolded glycoproteins from the ER in a proteasome-dependent manner (Li et al., 2006, 2008; Zhao et al., 2007). Later studies demonstrated that the human protein UBXD1 associates with p97 through PUB–PIM interactions (Madsen et al., 2008; Blueggel et al., 2019). In UBXD1, PUB is positioned at the center and forms intramolecular interactions with an extended UBX (eUBX) domain at the C-terminus of UBXD1. Notably, while maintaining the PIM motif-binding pocket on one side, the PUB–eUBX forms a unique module that binds ubiquitin from unfolded substrates exiting the p97 C-terminal pore (Blueggel et al., 2023). Similarly, ECLIPSE associates with the C-terminus of CDC48A and retains several key features of the PIM-binding motif (Fig. 3). However, ECLIPSE lacks a C-terminal eUBX domain, suggesting that while it can bind CDC48, it is not involved in the downstream processing of CDC48 substrates. Instead, PUX2, a member of the Arabidopsis PUX family that is structurally similar to UBXD1, contains both a central PUB domain with a conserved PIM-binding motif and an unusually long UBX domain at the C-terminus (Zhang et al., 2021). This structural similarity raises the possibility that PUX2 may function as a plant homolog of human UBXD1.

Structural characterization revealed that UBXD1 binds to both the N- and C-termini of p97 via the VCP/p97-interacting motif (VIM) and the PUB domain, respectively (Blueggel et al., 2023), and the VIM domain is located <100 amino acids away from the PUB domain. In contrast, ECLIPSE is unable to bind the N-terminus of CDC48. Compared with UBXD1, ECLIPSE has a significantly more extended N-terminus, which has acquired two zinc-finger domains and a UBA domain, suggesting a potential role in binding ubiquitin or ubiquitinated proteins. The ECLIPSE PUB domain is separated from the UBA domain by ~150 amino acids, mostly arranged in a predicted coiled-coil structure, implying that the UBA domain of ECLIPSE extends beyond the N-terminus of CDC48. This arrangement positions ECLIPSE as an ideal molecular bridge, facilitating CDC48 recruitment via PUB–PIM association and directing it towards ubiquitinated substrates through its N-terminal ubiquitin-binding domain (Fig. 6). Indeed, truncations of the ECLIPSE UBA domain disrupt its association with SUN1 at the INM (Fig. 4). We propose a model for SUN1 degradation at the INM, where ECLIPSE serves as an adaptor for CDC48, specifically targeting CDC48 to SUN1 for proteasome-dependent degradation (Fig. 6). It is noteworthy that the evolution of ECLIPSE as a CDC48 adaptor may be a feature specific to land plants because the ECLIPSE protein domain architecture appears highly conserved across land plants but absent in animals and algae (Fig. 2). This raises the intriguing possibility that the ECLIPSE-dependent INMAD plays a specific role in maintaining nuclear envelope protein homeostasis and providing unique advantages to terrestrial plant cells.

Fig. 6.

A working model for ECLIPSE-mediated SUN1 degradation, as part of the plant INMAD pathway. ECLIPSE serves as a potential adaptor in the INMAD pathway for SUN1 degradation (left). We propose that the N-terminal UBA domain of ECLIPSE recognizes polyubiquitinated SUN1 whereas its C-terminal PUB domain recruits the CDC48 complex to facilitate the retrotranslocation and subsequent degradation of polyubiquitinated SUN1. PUX5 acts as a negative regulator of the INMAD pathway, potentially binding ubiquitinated SUN1 via its UBA domain to inhibit elongation of the ubiquitin chain (right). In addition, PUX5 binds the N-terminus of CDC48 via its UBX domain. These interactions are possible to compete with ECLIPSE for binding with both SUN1 and CDC48, thereby protecting SUN1 from degradation.

SUN1 degradation represents the first instance of INMAD characterized in plants, and PUX3/4/5 are the only CDC48 cofactors previously reported to function in plant INMAD. In contrast to ECLIPSE, they redundantly protect SUN1 from UPS-dependent degradation, despite their association with CDC48 (Huang et al., 2020). While the exact mechanism behind this protective function remains unclear, PUX5, like ECLIPSE, also contains a UBA domain. The UBA domain presents an interesting dichotomy in cellular function, as it can promote either protein degradation or protein stabilization. UBA domains involved in degradation typically facilitate the transfer of polyubiquitinated substrates to the proteasome, whereas those that stabilize proteins prevent ubiquitin chain elongation by capping the substrate-linked ubiquitin. Our data indicate that the UBA domain of ECLIPSE is essential for its interaction with SUN1 (Fig. 4) and that ECLIPSE positively regulates SUN1 degradation (Fig. 5), suggesting that its UBA domain may facilitate the binding of ubiquitinated SUN1 for degradation. In contrast, while it remains unclear whether the UBA domain of PUX5 is necessary for its interaction with SUN1, it may play a stabilizing role, supported by the negative regulatory function of PUX5 in SUN1 degradation (Huang et al., 2020). A possible regulatory mechanism may involve competitive binding of ubiquitinated substrates by the UBA domains of ECLIPSE and PUX5, fine-tuning INM protein degradation in plants: PUX5 binding to ubiquitinated SUN1 could potentially inhibit the elongation of the ubiquitin chain, blocking ECLIPSE from binding and thereby protecting SUN1 from degradation. Additionally, PUX5 contains a UBX domain, which also interacts with CDC48. As a result, perinuclear CDC48 contains two modules that dynamically and competitively interact with ECLIPSE and PUX5. It is tempting to speculate that these interactions constitute a flexible mechanism for regulating INM substrate degradation (Fig. 6).

Supplementary data

The following supplementary data are available at JXB online.

Table S1. Primers used in this study.

Table S2. Co-expression analysis between ECLIPSE and other genes.

Glossary

Abbreviations:

INM

inner nuclear membrane

INMAD

INM-associated degradation

UPS

ubiquitin–proteasome system

UBX

UBIQUITIN REGULATORY X

PUX

PLANT UBX DOMAIN-CONTAINING

Author contribution

EC, MJ, and YG: designed the research; EC and MJ: performed immunoblotting; EC and OX: performed Y2H analysis; EC: performed all other experiments. All authors analyzed the data, discussed the results, and edited the manuscript.

Conflict of interests

The authors declare no conflict of interests.

Funding

This work was supported by the U.S. National Science Foundation (MCB-2049931), the National Institute of Food and Agriculture (HATCH project CA-B-PLB-0352-H), and the National Institute of General Medical Sciences of the National Institutes of Health (1R35GM154623-01) (to YG).

Data availability

Raw data files for mass spectrometry reanalysis are publicly available at the ProteomeXchange Consortium via the PRIDE partner repository (identifier: PXD015919 and PXD015920).