Proximity labeling proteomics reveals critical regulators for inner nuclear membrane protein degradation in plants

ABSTRACT The inner nuclear membrane (INM) selectively accumulates proteins that are essential for nuclear functions; however, overaccumulation of INM proteins results in a range of rare

genetic disorders. So far, little is known about how defective, mislocalized, or abnormally accumulated membrane proteins are actively removed from the INM, especially in plants and animals.

Here, via analysis of a proximity-labeling proteomic profile of INM-associated proteins in _Arabidopsis_, we identify critical components for an INM protein degradation pathway. We show

that this pathway relies on the CDC48 complex for INM protein extraction and 26S proteasome for subsequent protein degradation. Moreover, we show that CDC48 at the INM may be regulated by a

subgroup of PUX proteins, which determine the substrate specificity or affect the ATPase activity of CDC48. These PUX proteins specifically associate with the nucleoskeleton underneath the

INM and physically interact with CDC48 proteins to negatively regulate INM protein degradation in plants. SIMILAR CONTENT BEING VIEWED BY OTHERS GLOBAL PROFILING OF PLANT NUCLEAR MEMBRANE

PROTEOME IN _ARABIDOPSIS_ Article 29 June 2020 ADAPTOR PROTEIN COMPLEX INTERACTION MAP IN _ARABIDOPSIS_ IDENTIFIES P34 AS A COMMON STABILITY REGULATOR Article 12 January 2023 CARGO SORTING

ZONES IN THE _TRANS_-GOLGI NETWORK VISUALIZED BY SUPER-RESOLUTION CONFOCAL LIVE IMAGING MICROSCOPY IN PLANTS Article Open access 26 March 2021 INTRODUCTION The nuclear envelope (NE) is

composed of two concentric lipid bilayers, the outer nuclear membrane (ONM) that is continuous with the endoplasmic reticulum (ER), and the inner nuclear membrane (INM) that faces the

nucleoplasm. The ONM is jointed with the INM at the nuclear pore, where the nuclear pore complex (NPC) resides. As a distinct membrane territory, the INM hosts a unique set of membrane

proteins that are required for essential nuclear functions, such as genome organization, transcriptional control, mechanosensation, and signal transduction1. However, abnormal accumulation

of INM proteins has been associated with altered nuclear morphology and a range of rare genetic diseases2,3. For example, overaccumulation of the evolutionarily conserved INM protein SUN1

(Sad1 and UNC84 Domain Containing 1) is pathogenic in humans and is linked to muscular dystrophy and premature aging syndrome4. Despite the obvious importance in maintaining protein

homeostasis and integrity at the INM, we know very little about the mechanism that removes abnormally accumulated or defective membrane components from the INM, especially in plants and

animals. Previous studies in yeasts have uncovered an INM-associated degradation (INMAD) pathway that mediates destruction of integral INM proteins5,6. Nevertheless, key components in yeast

INMAD pathway are missing in plants and animals. For example, the initiation of yeast INMAD relies on membrane-bound E3 ubiquitin ligases Asi1 and Asi3, which localize to the INM and

ubiquitinate misfolded or mislocalized INM proteins for proteasomal degradation5,6. However, Asi genes do not have homologs in plants and animals. Doa10 is another E3 ligase found in ER and

INM and participates in both ER-associated degradation (ERAD) and INMAD in yeasts7. In addition, the Anaphase-promoting complex (APC/C) mediates the degradation of INM protein Mps3 in

yeasts8. Nonetheless, neither Doa10 nor APC/C homolog has been linked to INMAD in plants or animals. Recently, it was reported that the human INM protein Emerin is selectively cleared under

ER stress through vesicular transport to lysosomes, but this process depends on Emerin’s LEM domain and is not observed for other INM proteins9. So far, we still miss key evidence to support

the existence of a ubiquitin/proteasome-dependent INMAD pathway outside of yeasts. Proximity-labeling-based proteomic approaches have been recently applied to plant research and have been

shown to be invaluable in profiling functional components of various protein complexes10,11,12,13,14. By combining the power of BioID2-based proximity labeling (PL)15, label-free

quantitative mass spectrometry (LFQMS), and ratiometric analysis (RA), we profiled the NE-associated proteome in _Arabidopsis_ using known NE proteins as bait16. Here, we report a group of

periphery NE proteins that are explicitly associated with the INM and involved in ubiquitin-mediated proteolysis. They include CDC48 proteins, its cofactors UFD1 and NPL4, and a specific

subgroup of plant ubiquitin regulatory X (UBX) domain-containing proteins (PUXs). CDC48, also known as p97 in mammals, is a conserved AAA-ATPase molecular chaperon that can mediate the

extraction of integral proteins from the membrane and recruit 26S proteasome for subsequent protein degradation17,18. PUX proteins define a plant protein family that possesses a conserved

UBX domain, which mediates direct interactions with CDC48 proteins19. Some PUXs also contain a ubiquitin-associated (UBA) domain, which allows them to bind ubiquitinated protein substrates

and act as selective adapters between CDC48 and membrane substrates20,21,22. However, PUX without the UBA domain has been shown to directly interfere with the CDC48 activity23,24,25. We used

the conserved INM protein SUN1 as a model to dissect the functional connection between the INM-CDC48 association and INM protein degradation in plants. We showed that SUN1 undergoes

constitutive degradation in a proteasome-dependent and autophagy-independent manner in _Arabidopsis_, and this process engages the CDC48 complex. Moreover, we showed that the CDC48 activity

at the INM may be directly regulated by a specific subgroup of PUX proteins, including PUX3, PUX4, and PUX5. These PUX proteins have evolved a membrane preference for the INM through

association with the nucleoskeleton and physically interact with CDC48 proteins to negatively regulate INM protein degradation in plants. RESULTS IDENTIFICATION OF PROTEINS SPECIFICALLY

ASSOCIATED WITH THE INM Previously we fused a promiscuous biotin ligase (BioID2) to AtSUN1, a conserved and one of the best characterized integral INM resident protein, and performed

PL-LFQMS experiment using a _35__S: HA-BioID2-SUN1_ transgenic line to profile INM-associated proteins16. Here, we performed additional validation, including the relative protein expression

level, NE localization, and inducible biotinylation of HA-BioID2-SUN1 in the transgenic line to further support the specificity and efficiency of our previous PL-LFQMS profiling

(Supplementary Fig. 1 and Supplementary Note 1). To identify candidates that are specifically associated with SUN1 at the INM and exclude those that associate with both the INM and the ONM,

we reanalyzed the SUN1 PL-LFQMS data using the ONM-anchored protein WIT126 as a control (Fig. 1a). We generated transgenic plants expressing BioID2-tagged WIT1 and performed PL-LFQMS. The

resulting MS data together with the available MS data from BioID2-SUN1 and plants expressing YFP-BioID2 without biotin treatment (Mock)16 were used for a three-dimensional ratiometric

analysis. The peptide intensities were normalized across SUN1, WIT1, and Mock samples and were used to perform pairwise comparisons (Fig. 1b). Using _p_-value < 0.05 and fold-change >

2 as cutoffs for both controls, we obtained 15 SUN1-specific preys (Fig. 1c and Supplementary Fig. 2a). Among them, five are components of the plant nucleoskeleton, including CRWN1, CRWN2,

CRWN3, CRWN4, and KAKU4, consistent with the well-established interaction between SUN1 and the nucleoskeleton at the INM27,28. On the other hand, we identified a total of four WIT1-specific

preys, including WIP1, WIP3, RanGAP1, and RanGAP2 (Fig. 1c and Supplementary Fig. 2b), supporting previous reports that WIT1 forms complexes with WIP proteins and anchors RanGAPs to the

ONM26,29,30. These analyses reinforce the high specificity of PL-LFQMS-RA in identifying local proteome in plants16. An improved biotin ligase enzyme with higher labeling efficiency termed

TurboID has been recently developed13,14,31. Using TurboID in PL-LFQMS-RA may further improve its efficiency and result in a more comprehensive profiling. THE CDC48 COMPLEX ASSOCIATES WITH

INTEGRAL INM PROTEINS Intriguingly, among the SUN1-specific preys, four proteins are potentially associated with the molecular chaperone CDC48, including two ubiquitin fusion degradation

(UFD) proteins (UFD1B and UFD1C) and two PUX proteins (PUX4 and PUX5) (Fig. 1c). In yeasts and mammals, UFD proteins are part of a CDC48-UFD-NPL4 heterotrimeric complex, which mediates the

membrane extraction of ubiquitinated proteins and recruits 26S proteasome for subsequent protein degradation17,18. PUX proteins also can directly interact with CDC48 and function to regulate

CDC48 activity or to bridge CDC48 with ubiquitinated membrane protein substrates20,23,24,25,32,33. To verify the connection of the identified UFD and PUX proteins with CDC48, we performed

yeast-two-hybrid experiments. We found that PUX4, PUX5, UFD1B, and UFD1C interacted with three _Arabidopsis_ CDC48 paralogs, except that the interaction between UFD1B and CDC48C might be

weak (Fig. 1d). This result supports the idea that PUX4, PUX5, UFD1B, and UFD1C may directly participate in the CDC48-dependent proteolysis pathway in plants. Notably, when we performed

reanalyses on a series of previously published PL-LFQMS profiling datasets using NE proteins as baits16, we found that UFD1B, UFD1C, PUX4, and PUX5 were also probed by another INM protein

NEAP1 (Fig. 1e and Supplementary Fig. 2c). In contrast, none of these UFD and PUX proteins were detected by other baits, including the ONM protein WIP1 and WIT1, the NPC component Nup93a and

Nup82, and the ER membrane protein RHD3. In addition, PL-LFQMS profiling was performed using a third INM bait protein NEMP_A, an _Arabidopsis_ ortholog of the Xenopus INM protein Nemp1, and

PUX4 and PUX5 were also probed (Fig. 1e and Supplementary Fig. 2c). These data are consistent with the idea that there is a specific association of INM proteins with the identified

CDC48-dependent membrane protein degradation machinery. INM PROTEIN SUN1 UNDERGOES PROTEASOME-DEPENDENT DEGRADATION The degradation of INM proteins and associated regulatory mechanisms have

not been defined in plants or metazoans. To gain functional insight into the association of INM proteins with UFD and PUX proteins and its potential connection with the INMAD, we first

assayed whether INM protein homeostasis involves proteasomal degradation in plants. We found that treatment of _35__S: HA-BioID2-SUN1_ seedlings with the proteasome inhibitor MG132

significantly increased the steady-state accumulation of SUN1 (Fig. 2a). Moreover, MG132 treatment also induced detectable accumulation of polyubiquitinated SUN1 (Fig. 2b), suggesting that

SUN1 may undergo ubiquitin/proteasome-dependent degradation. Next, we set up an in vivo protein degradation assay. We found that when the _35__S: HA-BioID2-SUN1_ seedlings were treated with

cycloheximide (CHX), which blocks the de novo protein synthesis, the level of SUN1 decayed with time, suggesting that the degradation is constitutive (Fig. 2c). Adding MG132 to CHX-treated

plants significantly inhibited SUN1 degradation over time (Fig. 2d), whereas adding concanamycin A to block autophagy did not affect SUN1 degradation (Fig. 2e). These results suggest that at

least some INM proteins are degraded through a proteasome-dependent but autophagy-independent degradation pathway in plants. THE ENTIRE CDC48 COMPLEX WAS CAPTURED BY SUN1 Knowing that the

proteasome participates in SUN1 degradation, we performed a second round of PL-LFQMS profiling using MG132-treated _HA-BioID2-SUN1_ plants to trap the degradation machinery. Remarkably,

MG132 treatment enabled HA-BioID2-SUN1 to capture all core components of the CDC48-UFD-NPL4 trimeric complex, including CDC48B, CDC48C, UFD1B and UFD1C, and NLP4A (Fig. 2f and Supplementary

Fig. 3a). This result strongly supports the involvement of the CDC48 complex in the proteasome-dependent INM protein degradation. PUX3/4/5 SELECTIVELY ASSOCIATE WITH THE INM Beside the

entire CDC48 complex, MG132 treatment allowed HA-BioID2-SUN1 to capture a third PUX protein (PUX3) in addition to PUX4 and PUX5 (Fig. 2f and Supplementary Fig. 3a). Intriguingly, PUX3, PUX4,

and PUX5 are closely related and form a subclade within the 16 PUXs encoded by _Arabidopsis_ (Fig. 2g and Supplementary Fig. 3b). Emerging evidence suggests that the 16 AtPUX proteins play

an important role in determining the substrate specificity and ATPase activity of CDC48 and are functionally diversified by associating with different membrane compartments. For example,

PUX10 explicitly localizes to the lipid droplet (LD) membrane via a unique hydrophobic polypeptide sequence and recruits CDC48 for LD protein degradation21,22. PUX7, PUX8, PUX9, and PUX13

associate with the autophagosome by interacting with ATG8 through their ubiquitin-interacting motif (UIM)-like sequence and recruit inactive CDC48 molecules for degradation34. In our

PL-LFQMS experiments, INM baits only probed PUX members in the PUX3/4/5 subclade (Fig. 3a and Supplementary Fig. 4a). Because these _PUX_ genes are not expressed at a higher level than other

_PUX_s (Supplementary Fig. 4b), identification of PUX3/4/5 suggests their specific function at the INM. In line with this observation, the PUX3/4/5 subgroup is homologous to the yeast UBX

protein UBX1 (Supplementary Fig. 3b), which is involved in the degradation of yeast INM protein Asi135. To confirm the specific association of PUX3/4/5 with the INM, we transiently

coexpressed SUN1 with PUX4, PUX5, and PUX7, respectively. Co-immunoprecipitation (co-IP) assay showed that SUN1 was specifically co-purified with PUX4 and PUX5, but not PUX7 (Fig. 3b). We

also examined the association of PUX4 with proteins from different membrane compartments. We transiently coexpressed PUX4 with four membrane proteins, including SUN1 at the INM, WIT1 at the

ONM, BRI1 at the plasma membrane, and ARA6 at the late endosome. Again, the co-IP result demonstrated that PUX4 specifically pulled down SUN1 but not the other membrane proteins tested (Fig.

 3c). However, when we used bimolecular fluorescence complementation (BiFC) assay to verify the interaction between SUN1 and PUX4, only very weak signal was detected (Supplementary Fig. 4c),

suggesting that PUX3/4/5 may not interact with INM proteins constitutively but in a transient or indirect manner. Nevertheless, these data further support the specific association of PUX3,

PUX4, and PUX5 with the INM. PUX3/4/5 INTERACT WITH THE NUCLEOSKELETON To investigate how PUX3/4/5 are selectively associated with the INM, we generated _35__S: PUX5-BioID2_ transgenic

plants and performed PL-LFQMS profiling. Among the top candidates, we identified both PUX3 and PUX4 but no other PUXs, suggesting that PUX3, PUX4, and PUX5 may function together at the INM

(Fig. 3d). We also identified components of the CDC48 complex, including CDC48A, CDC48B, UFD1B, UFD1C, and NPL4A, reinforcing the direct interaction between PUX3/4/5 and the CDC48 complex

(Fig. 1d). Importantly, CRWN4, a component of the nucleoskeleton, was identified as a specific PUX5 interactor with high confidence (Fig. 3d). To confirm the interaction of PUX3, PUX4, and

PUX5 with the nucleoskeleton, we performed BiFC assay between PUX3/4/5 and another nucleoskeleton component KAKU4. Different from the BiFC signal between PUX4 and SUN1, which was only weakly

detected, robust complemented fluorescence was observed at the nuclear periphery between PUX3/4/5 and KAKU4 (Fig. 3e), suggesting a stable association of PUX3, PUX4, and PUX5 with the

nucleoskeleton. This data implicates a mechanism for the INM targeting of PUX3, PUX4, and PUX5 and explains their proximity with multiple INM proteins. PUX3/4/5 NEGATIVELY REGULATE SUN1

DEGRADATION To investigate the functional importance of PUX3, PUX4, and PUX5 in INM protein degradation, we took advantage of the CRISPR/Cas9 gene-editing tool and generated a series of

_pux_ mutants in an isogenic _35__S: HA-BioID2-SUN1_ background, including a _pux3_ single mutant, a _pux3 pux4_ and a _pux3 pux5_ double mutant, and a _pux3 pux4 pux5_ triple mutant. The

Cas9-generated mutations were all nonsense and occurred near the beginning of target genes, which led to frameshifts and premature stop codons, likely resulting in loss-of-function mutations

(Fig. 4a and Supplementary Fig. 5a). Next, we treated _HA-BioID2-SUN1_ seedlings with CHX in WT and _pux_ mutant background in parallel to compare the turnover rate of SUN1 protein.

Surprisingly, we found that the SUN1 turnover rate is faster in the _pux3_ single mutant than that in WT, although SUN1 steady-state accumulation is unchanged. This effect was further

enhanced in the _pux3 pux4_ and _pux3 pux5_ double mutants (Supplementary Fig. 5b). The accelerated turnover rate was most dramatic in the _pux3 pux4 pux5_ triple mutant, which showed a

nearly 70% reduction in SUN1 protein level within 4 h of CHX treatment (Fig. 4b, c). Nevertheless, MG132 treatment greatly compromised the SUN1 degradation in the _pux3 pux4 pux5_ triple

mutant (Supplementary Fig. 5c), similar to what was observed in WT plants (Fig. 2d), suggesting that the regulation by PUX3/4/5 is upstream of the proteasome. Consistent with the enhanced

degradation of SUN1 in the absence of PUX3, PUX4, and PUX5, when PL-LFQMS was performed using _HA-BioID2-SUN1_ plants in WT and _pux3 pux4 pux5_ mutant background, we observed a significant

decrease in the biotinylation level of the nucleoskeleton, including CRWN1, CRWN2, and CRWN3, in the _pux3 pux4 pux5_ mutant compared to WT (Fig. 4d). Because that the steady-state SUN1

accumulation was not obviously affected in the triple mutant, the reduced capacity of SUN1 in labeling its proximal proteins may attribute to its reduced stability and half-life in the INM

in the absence of PUX3, PUX4, and PUX5. Together, these data demonstrated that PUX3, PUX4, and PUX5 are redundantly required for the protection of INM protein from proteasomal degradation.

DISCUSSION The membrane-associated protein degradation mechanisms are essential for the maintenance of integrity and identity of membrane organelles in eukaryotes and have been under

extensive investigation. In addition to the conserved and well-characterized ERAD pathway36,37,38, chloroplast-associated protein degradation system (CHLORAD)39 and lipid droplet-associated

degradation (LDAD) pathway21,22 were recently discovered in plants. Here we report the identification of critical components for a ubiquitin/proteasome-dependent INM protein degradation

pathway in plants using proximity-labeling proteomics. This pathway raises an important parallel with ERAD, CHLORAD, and LDAD pathways, all of which exploit the CDC48 complex. It supports a

converging theme that the ATP-driven chaperon plays a central role in mediating membrane-associated protein degradation at different organelles. Consistently, the CDC48 protein is wildly

distributed throughout the cell, and the functionally characterized CDC48 ortholog in _Arabidopsis_ (CDC48A) is essential for plant survival40. Although CDC48 is the common executor for the

membrane-associated protein destruction, the substrate specificity and ATPase activity of CDC48 is subject to organelle-specific regulations. In particular, the functionally diversified PUX

proteins appear to play an important role in this process. PUX10 has been reported to localize to lipid droplets and bridge CDC48 with membrane substrates in lipid droplets32,33. In

contrast, PUX7/8/9/13 were shown to associate with autophagosomes and recruit inactive CDC48 for degradation34. Furthermore, PUX1 has been demonstrated to disrupt the CDC48 activity by

directly promoting dissociation of the CDC48 hexamer23,24. Here, we found that another PUX subgroup (PUX3/4/5) is uniquely recruited to the INM through interacting with the nucleoskeleton

underneath the INM. There, they negatively regulate the INM protein degradation by physically interacting with CDC48. We propose two possible models to explain the role of PUX3/4/5 during

this process, and the two models are not mutually exclusive. First, PUX3/4/5 may directly interfere with the CDC48 protein activity, like PUX1 (Fig. 4e). Because CDC48 was found highly

concentrated in the nucleus and the nuclear periphery in _Arabidopsis_40, PUX3/4/5 may function to inhibit the CDC48 activity at the INM and prevent non-specific degradation of INM proteins

until they are ubiquitinated and ready for turnover. Second, PUX3/4/5 may mediate the degradation of putative E3 ubiquitin ligases that initiate the INM protein degradation in plants (Fig. 

4e). In yeast, INM-associated E3 ligase Asi1 is critical for the ubiquitination of INM substrates for INMAD. The turnover of Asi1 itself depends on yeast UBX protein UBX1 and the CDC48

complex35. Because the PUX3/4/5 subclade bears homology with yeast UBX1 (Supplementary Fig. 3b), PUX3/4/5 may promote the stability of INM proteins through recruiting the CDC48 complex for

degradation of INM-associated E3 ligases in plants. However, whether plant INM protein degradation is initiated by INM-associated E3 ubiquitin ligases, and if so, what the E3 ligases are,

remain to be determined. Lastly, the association of PUX3/4/5 with CRWN4 and KAKU4 suggests a nucleoskeleton-embedded regulatory mechanism for INM protein degradation in plants. The

connection of the nucleoskeleton with the INM protein degradation is intriguing because it extends the functional importance of the nucleoskeleton to integral INM proteins, making it not

only a scaffold for binding and retention of INM proteins41 but also a platform that hosts molecules to regulate INM protein stability. METHODS PLANT GROWTH CONDITIONS _Arabidopsis_

(_Arabidopsis thaliana_) seeds were surface sterilized with 70% ethanol for 1 min and 20% (v/v) bleach for 8 min and rinsed with sterile water for five times before planted on the ½

Murashige and Skoog (½ MS) medium supplied with 1% agar. After 2 days of stratification at 4 °C, plants were moved to the plant growth room or chamber and grown under a photoperiod of 16 

h-light and 8 h-dark at 22 °C. For proximity labeling and mass spectrometry, 10-day-old seedlings were used. _Nicotiana benthamiana_ plants used for transient protein expression were grown

under the same conditions except that the seeds were planted directly into the soil. VECTOR CONSTRUCTION FOR PROXIMITY LABELING Cloning was performed using a multisite gateway system as

reported before42,43. Briefly, _HA-BioID2_ and _PUX5_ were cloned into the multisite gateway entry vector pBSDONR p1-p4 using BP reaction (BP clonase™ II, Thermo Fisher, Cat#11789020).

_BioID2-HA_, _WIT1_, _RHD3_, and _NEMP_A_ were cloned into pBSDONR p4r-p2. BioID2 fusions were generated by fusing constructs in pBSDONRp1-p4 (N-terminus) and pBSDONRp4r-p2 (C-terminus) into

the destination vector pEarlyGate100 by LR reaction (LR clonase™ II plus, Thermo Fisher, Cat#12538200). All primers used in this study were listed in Supplementary Data 1. PLANT MATERIALS

_Arabidopsis thaliana_ ecotype Columbia (Col-0) was used as the wild-type (WT). Transgenic _Arabidopsis_ lines used for proximity labeling (_35S__: HA-BioID2-WIT1, 35__S: HA-BioID2-NEMP_A,

35__S: HA-BioID2-RHD3_, and _35__S: PUX5-BioID2-HA_) were generated by floral dip transformation of WT plants with Agrobacterium strain GV3101 carrying corresponding constructs. T1

transgenic plants were selected by Basta resistance, and the protein expression level of individual lines was examined by immunoblotting using streptavidin-HRP (Abcam, Cat#7403, dilution

1:10,000) and anti-HA antibody (3F10, Roche, Cat#11867431001, dilution 1:5000) using T2 seedlings treated with 50 µM biotin. The _35__S: HA-BioID2-SUN1_ transgenic line was described

before16. PROXIMITY LABELING AND FREE BIOTIN DEPLETION For proximity labeling, 10-day-old transgenic seedlings expressing protein tagged with BioID2 were submerged in 50 µM biotin solution

for 16 h at room temperature before being washed three times with ice-cold water and harvested. Total protein was extracted by grinding the sample with extraction buffer [50 mM Tris-HCl (pH

7.5), 150 mM NaCl, 0.5% NP40 (v/v), 0.5% Triton X-100 (v/v), 0.5% sodium deoxycholate (w/v), 40 µM MG132, 1× Protease Inhibitor Cocktail]. The supernatant was collected after samples were

centrifuged at 13,000 rpm at 4 °C for 10 min. REMOVAL OF FREE BIOTIN BY DESALTING CHROMATOGRAPHY For desalting chromatography, the AKTA protein purification system (GE Healthcare) with 3 × 5

 mL HiTrap desalting columns (GE Healthcare, Cat#GE29-0486-84) was utilized. About 2 mL of total proteins for each sample were loaded into the HiTrap column after equilibration with

desalting buffer (50 mM Tris-HCl pH 7.5, 0.05% Triton X-100) using 5 mL syringe with a 2 mL sample loop. The flow rate used for all experiments was 0.7 mL/min with 0.3 Mpa as the pressure

limit. The salt elute containing free biotin with high conductivity were abandoned while the total protein elute with UV280 absorbance peak was collected for affinity purification (AP).

AFFINITY PURIFICATION BY STREPTAVIDIN-COATED BEADS About 3 mL desalted protein elute was collected for each sample and mixed with 50 μL streptavidin-coated magnetic beads (Dynabeads™ Myone™

Streptavidin C1, Thermo Fisher, Cat#65002) for AP. After incubation with rotation at 4 °C overnight, the beads were separated on a magnet rack and washed with extraction buffer for five

times. Proteins on beads were then eluted by boiling at 98 °C for 20 min in elution buffer [50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% SDS, 50 µM biotin] before separated by SDS-PAGE. Each

lane of SDS-PAGE gel was cut into four individual pieces and digested with trypsin in 50 mM ammonium bicarbonate at 37 °C overnight, extracted twice with 1% formic acid in 50% acetonitrile

aqueous solution, and dried by SpeedVac before LC-MS/MS analysis was performed. LC-MS/MS For LC-MS/MS analysis, tryptic peptides were separated by a 135 min gradient elution at a flow rate

0.300 µL/min with the DIONEX ultimate 3000 integrated nano-HPLC system (Thermo Fisher) which is directly interfaced with the Thermo LTQ-Orbitrap mass spectrometer. The analytical column was

Acclaim PopMap™ 100 C18 capillary column (75 µm ID, 150 mm length, Thermo Fisher Scientific, Cat#164568) packed with C-18 resin (300 Å, 5 µm, Varian). Mobile phase A consisted of 0.1% formic

acid, and mobile phase B consisted of 100% acetonitrile and 0.1% formic acid. The LTQ-Orbitrap mass spectrometer was operated in the data-dependent acquisition mode using the Xcalibur 4.1

software. There is a single full-scan mass spectrum in the Orbitrap (400–1800 _m_/_z_, 30,000 resolution) followed by 20 data-dependent MS/MS scans in the ion trap at 35% normalized

collision energy. MS/MS spectra from each LC-MS/MS run were searched against the TAIR10 database using Proteome Discoverer (Version 2.2; Thermo Fisher) with the following criteria: full

tryptic specificity was required; two missed cleavages were allowed; carbamidomethylation was set as fixed modification; oxidation (M) were set as variable modifications; precursor ion mass

tolerance was 20 ppm for all MS acquired in the Orbitrap mass analyzer; and fragment ion mass tolerance was 0.02 Da for all MS spectra. High confidence score filter (FDR < 1%) was used to

select the “hit” peptides and their corresponding MS/MS spectra were manually inspected. MS DATA ANALYSIS For proximity biotinylation data analysis, protein enrichment areas (LFQ values)

were integrated and used as the input for sample normalization by DEP package (version 1.8.0) in RStudio (version 1.1.463). Specifically, the output file from Proteome Discoverer (Version

2.2) was imported into DEP using the LFQ intensities as main category. The data matrix was filtered to remove proteins with missing values using less stringent filtering (thr = 1) and then

was background corrected and normalized by variance stabilizing transformation. The remaining missing values in the dataset were further imputed (fun = “MinProb”, _q_ = 0.01) for missing not

at random (MNAR), which indicate that proteins are below the detection limit in specific samples (e.g., in the control samples). Then differential enrichment analysis which is based on

linear models and empirical Bayes statistics was performed with manually specified contrasts. Significant proteins are determined by defined cutoffs using fold-change and _p_-vaule compared

to controls. YEAST TWO-HYBRID ASSAYS The CDS of _PUX4/5, CDC48A/B/C_, and _UFD1B/1__C_ were fused to the GAL4-activation domain or GAL4-binding domain and cloned into the pGBKT7 or pGADT7

vectors by seamless ligation with ClonExpress II One Step Cloning Kit (Vazyme, Cat#C112). All fusion constructs were confirmed by sequencing. These constructs were transformed into

_Saccharomyces cerevisiae_ strain Y187 and Y2HGold (Yeastmaker Yeast Transformation System 2, Clontech), and transformants were selected using colony PCR. Y187 and Y2HGold were mated in 4 mL

2× YPDA medium at 30 °C for 18–22 h. The resulting culture containing diploid yeasts was diluted and dropped on DDO medium (SD-Leu/-Trp) and QDO medium (SD-Leu/-Trp/-His/-Ade) and incubated

at 30 °C for 3 days before photos were taken. Autoactivation assays have been performed for each bait and prey construct with corresponding empty vector to exclude potential false

positives. IN VIVO PROTEIN DEGRADATION ASSAY Ten-day-old seedlings were treated by 100 µM CHX with or without 50 µM MG132 or 1 µM Concanamycin A for the indicated time. About 20 seedlings

were sampled per treatment per time point. The samples were dried with paper towels and weighed to keep the biomass the same for each sample. Samples were then frozen in liquid nitrogen and

ground to fine powder. Protein was extracted by adding 200 µL extraction buffer [50 mM Tris-HCl (pH7.5), 150 mM NaCl, 0.5% NP40 (v/v), 0.5% Triton X-100 (v/v), 0.5% sodium deoxycholate

(w/v), 40 µM MG132, 1× Protease Inhibitor Cocktail] to the fine powder and vortex mixing. The supernatant was collected after centrifuged at 13,000 rpm at 4 °C for 10 min. Protein samples

were adjusted to a final concentration of 300 mg/mL, separated by SDS-PAGE, and subject to immunoblotting using anti-HA (3F10, Roche, Cat#11867431001, dilution 1: 5000) and anti-Actin

antibody (Abiocode, Cat#R3772-1P, dilution 1: 3000). CRISPR/CAS9-MEDIATED KNOCKOUT OF _PUX3/4/5_ The single guide RNA (sgRNA) sequences that target _PUX3_, _PUX4_, and _PUX5_ were designed

using the webserver CRISPOR (http://crispor.tefor.net/). Using pCBC-DT1T2 as the template, the sgRNA (_PUX4_)-U6-26t-U6-29p-sgRNA (_PUX5_) cassette was amplified by PCR and inserted into

pHEE401 by GoldenGate assembly to obtain pHEE401-_PUX4_/_PUX5_. The sgRNA (_PUX3_) was cloned into pHEE401, which was then digested with BsaI to obtain the U6-26p-sgRNA (PUX3)-U6-29t

fragment. This fragment was then introduced into the pHEE401-_PUX4_/_PUX5_ by seamless cloning to obtain pHEE401-_PUX3_/_PUX4_/_PUX5_ construct. Another plasmid with a different set of

sgRNAs for _PUX3_/_PUX4_/_PUX5_ was constructed in a similar manner. The two pHEE401-_PUX3_/_PUX4_/_PUX5_ plasmids were transformed into the homozygous T3 lines of _35__S: HA-BioID2-SUN1_.

Mutants that contain different combinations of _PUX3_/_PUX4_/_PUX5_ mutations were identified in the T1 generation by amplifying and sequencing of the genomic fragments of

_PUX3_/_PUX4_/_PUX5_ genes. Homozygous mutations were obtained and confirmed by sequencing in the T2 generation. CO-IMMUNOPRECIPITATION AND BIFC Co-IP and BiFC experiments were performed

using transient protein expression in _N. benthamiana_ leaves infiltrated with Agrobacterium carrying corresponding constructs as described previously44. For co-IP, a 15 µL volume of HA

antibody (3F10, Roche) was added to 1 mL of total protein extract and incubated at 4 °C for 4 h with gentle shaking before 20 μL of pre-rinsed protein G agarose beads (Millipore) was added.

The mixture was incubated for another 3 h at 4 °C. The beads were washed five times with PBS buffer, separated by SDS-PAGE, and subjected to immunoblotting using anti-GFP (Living Colors,

Takara, Cat#632375, dilution 1:3000), anti-HA antibody (3F10, Roche, Cat#11867431001, dilution 1:5000), or anti-ubiquitin antibody (D9D5, Cell Signaling Technology, Cat#8081S, dilution

1:1000). For BiFC, leaves were collected for microscopic imaging 48 h after Agrobacterium infiltration. Confocal laser scanning microscopy was performed using a TCS SP8 STED confocal

microscope (Leica). sYFP fluorescence was excited by the 514 nm Argon laser and detected using a custom 522 nm to 545 nm band-pass emission filter, whereas mCherry fluorescence was excited

by the 561 nm laser and detected using a custom 595–620 nm band-pass emission filter. IMMUNOGOLD LABELING AND ELECTRON MICROSCOPY Roots of homozygous 35S: _HA-BioID2-SUN1_ T3 transgenic

plants were excised and immersed in 20% (w/v) BSA and frozen immediately in a high-pressure freezer (HPM100, Leica). Samples were put into tubes containing 0.2% uranyl acetate in acetone.

Freeze substitution and low-temperature embedding were carried out using Leica EM AFS2 as follows: −90 °C for 72 h, 2 °C/h increase for 15 h, −60 °C for 8 h, 2 °C/h increase for 15 h, −30 °C

for 8 h, and 2 °C per hour increase for 5 h to −20 °C. Samples were then washed 3 times with 100% ethanol for 3 h, infiltrated stepwise over 3 days at −20 °C in LR Gold resin and embedded

in capsules. The polymerization was performed at −20 °C for 24 h and room temperature for 3 days using UV light. Ultrathin sections were made using an ultramicrotome (Leica EM UC6). Samples

were treated with anti-HA primary antibody (3F10, Roche, Cat#11867431001, dilution 1:10) and 12 nm gold-conjugated goat anti-rat IgG (1:20, Jackson ImmunoResearch) on the grids and

post-stained in uranyl acetate for 30 min and in lead citrate for 3 min. Grids were imaged with a transmission electron microscope (H-7650B, Hitachi). REPORTING SUMMARY Further information

on research design is available in the Nature Research Reporting Summary linked to this article. DATA AVAILABILITY All data generated in this study has been made available either in the

Source Data file, via respective repository entry, or Supplementary Information files and are available from the corresponding author on reasonable request. Relevant mass spectrometry (MS)

proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository (Identifier: PXD015920). The MS raw data for BioID2-SUN1, BioID2-NEAP1, Nup93a-BioID2,

Nup82-BioID2, BioID2-WIP1, and Mock control were retrieved from PXD01591916. The MS datasets and figures associated with each dataset are listed in Supplementary Data 2. Differential

analysis and normalized MS data for BioID2-SUN1, BioID2-WIT1, and PUX5-BioID2 proximity-labeling proteomic experiments were provided in Supplementary Data 3. REFERENCES * Wilson, K. L. &

Berk, J. M. The nuclear envelope at a glance. _J. Cell Sci._ 123, 1973–1978 (2010). Article  CAS  Google Scholar  * Burke, B. & Stewart, C. L. Functional architecture of the cell’s

nucleus in development, aging, and disease. _Curr. Top. Dev. Biol._ 109, 1–52 (2014). Article  Google Scholar  * Jevtic, P., Edens, L. J., Vukovic, L. D. & Levy, D. L. Sizing and shaping

the nucleus: mechanisms and significance. _Curr. Opin. Cell Biol._ 28, 16–27 (2014). Article  CAS  Google Scholar  * Chen, C. Y. et al. Accumulation of the inner nuclear envelope protein

Sun1 is pathogenic in progeric and dystrophic laminopathies. _Cell_ 149, 565–577 (2012). Article  CAS  Google Scholar  * Foresti, O., Rodriguez-Vaello, V., Funaya, C. & Carvalho, P.

Quality control of inner nuclear membrane proteins by the Asi complex. _Science_ 346, 751–755 (2014). Article  ADS  CAS  Google Scholar  * Khmelinskii, A. et al. Protein quality control at

the inner nuclear membrane. _Nature_ 516, 410–413 (2014). Article  ADS  CAS  Google Scholar  * Koch, B. & Yu, H. G. Regulation of inner nuclear membrane associated protein degradation.

_Nucleus_ 10, 169–180 (2019). Article  CAS  Google Scholar  * Koch, B. A., Jin, H., Tomko, R. J. Jr. & Yu, H. G. The anaphase-promoting complex regulates the degradation of the inner

nuclear membrane protein Mps3. _J. Cell Biol._ 218, 839–854 (2019). Article  CAS  Google Scholar  * Buchwalter, A., Schulte, R., Tsai, H., Capitanio, J. & Hetzer, M. Selective clearance

of the inner nuclear membrane protein emerin by vesicular transport during ER stress. _eLife_ https://doi.org/10.7554/eLife.49796 (2019). * Lin, Q. et al. Screening of proximal and

interacting proteins in rice protoplasts by proximity-dependent biotinylation. _Front. Plant Sci._ 8, 749 (2017). Article  Google Scholar  * Khan, M., Youn, J. Y., Gingras, A. C.,

Subramaniam, R. & Desveaux, D. In planta proximity dependent biotin identification (BioID). _Sci. Rep._ 8, 9212 (2018). Article  ADS  Google Scholar  * Conlan, B., Stoll, T., Gorman, J.

J., Saur, I. & Rathjen, J. P. Development of a rapid in planta BioID system as a probe for plasma membrane-associated immunity proteins. _Front. Plant Sci._ 9, 1882 (2018). Article 

Google Scholar  * Mair, A., Xu, S. L., Branon, T. C., Ting, A. Y. & Bergmann, D. C. Proximity labeling of protein complexes and cell-type-specific organellar proteomes in _Arabidopsis_

enabled by TurboID. _eLife_ https://doi.org/10.7554/eLife.47864 (2019). * Zhang, Y. et al. TurboID-based proximity labeling reveals that UBR7 is a regulator of N NLR immune receptor-mediated

immunity. _Nat. Commun._ 10, 3252 (2019). Article  ADS  CAS  Google Scholar  * Kim, D. I. et al. An improved smaller biotin ligase for BioID proximity labeling. _Mol. Biol. Cell_ 27,

1188–1196 (2016). Article  CAS  Google Scholar  * Tang, Y., Huang, A. & Gu, Y. Global profiling of plant nuclear membrane proteome in Arabidopsis. _Nat. Plants_ 

https://doi.org/10.1038/s41477-020-0700-9 (2020). * Bays, N. W. & Hampton, R. Y. Cdc48-Ufd1-Npl4: stuck in the middle with Ub. _Curr. Biol._ 12, R366–R371 (2002). Article  CAS  Google

Scholar  * Ye, Y., Meyer, H. H. & Rapoport, T. A. The AAA ATPase Cdc48/p97 and its partners transport proteins from the ER into the cytosol. _Nature_ 414, 652–656 (2001). Article  ADS 

CAS  Google Scholar  * Schuberth, C. & Buchberger, A. UBX domain proteins: major regulators of the AAA ATPase Cdc48/p97. _Cell Mol. Life Sci._ 65, 2360–2371 (2008). Article  CAS  Google

Scholar  * Gallois, J.-L. et al. Functional characterization of the plant ubiquitin regulatory X (UBX) domain-containing protein AtPUX7 in _Arabidopsis thaliana_. _Gene_ 526, 299–308 (2013).

Article  CAS  Google Scholar  * Deruyffelaere, C. et al. PUX10 is a CDC48A adaptor protein that regulates the extraction of ubiquitinated oleosins from seed lipid droplets in _Arabidopsis_.

_Plant Cell_ 30, 2116 (2018). Article  CAS  Google Scholar  * Kretzschmar, F. K. et al. PUX10 is a lipid droplet-localized scaffold protein that interacts with cell division cycle48 and is

involved in the degradation of lipid droplet proteins. _Plant Cell_ 30, 2137 (2018). Article  CAS  Google Scholar  * Park, S., Rancour, D. M. & Bednarek, S. Y. Protein domain-domain

interactions and requirements for the negative regulation of _Arabidopsis_ CDC48/p97 by the plant ubiquitin regulatory X (UBX) domain-containing protein, PUX1. _J. Biol. Chem._ 282,

5217–5224 (2007). Article  CAS  Google Scholar  * Rancour, D. M., Park, S., Knight, S. D. & Bednarek, S. Y. Plant UBX domain-containing protein 1, PUX1, regulates the oligomeric

structure and activity of _Arabidopsis_ CDC48. _J. Biol. Chem._ 279, 54264–54274 (2004). Article  CAS  Google Scholar  * Arumughan, A. et al. Quantitative interaction mapping reveals an

extended UBX domain in ASPL that disrupts functional p97 hexamers. _Nat. Commun._ 7, 13047 (2016). Article  ADS  CAS  Google Scholar  * Zhao, Q., Brkljacic, J. & Meier, I. Two distinct

interacting classes of nuclear envelope-associated coiled-coil proteins are required for the tissue-specific nuclear envelope targeting of _Arabidopsis_ RanGAP. _Plant Cell_ 20, 1639–1651

(2008). Article  CAS  Google Scholar  * Haque, F. et al. SUN1 interacts with nuclear lamin A and cytoplasmic nesprins to provide a physical connection between the nuclear lamina and the

cytoskeleton. _Mol. Cell. Biol._ 26, 3738 (2006). Article  CAS  Google Scholar  * Haque, F. et al. Mammalian SUN protein interaction networks at the inner nuclear membrane and their role in

laminopathy disease processes. _J. Biol. Chem._ 285, 3487–3498 (2010). Article  CAS  Google Scholar  * Brkljacic, J., Zhao, Q. & Meier, I. WPP-domain proteins mimic the activity of the

HSC70-1 chaperone in preventing mistargeting of RanGAP1-anchoring protein WIT1. _Plant Physiol._ 151, 142–154 (2009). Article  CAS  Google Scholar  * Tatout, C., Evans, D. E., Vanrobays, E.,

Probst, A. V. & Graumann, K. The plant LINC complex at the nuclear envelope. _Chromosome Res._ 22, 241–252 (2014). Article  CAS  Google Scholar  * Branon, T. C. et al. Efficient

proximity labeling in living cells and organisms with TurboID. _Nat. Biotechnol._ 36, 880–887 (2018). Article  CAS  Google Scholar  * Deruyffelaere, C. et al. PUX10 is a CDC48A adaptor

protein that regulates the extraction of ubiquitinated oleosins from seed lipid droplets in _Arabidopsis_. _Plant Cell_ 30, 2116–2136 (2018). Article  CAS  Google Scholar  * Kretzschmar, F.

K. et al. PUX10 is a lipid droplet-localized scaffold protein that interacts with cell division cycle48 and is involved in the degradation of lipid droplet proteins. _Plant Cell_ 30,

2137–2160 (2018). Article  CAS  Google Scholar  * Marshall, R. S., Hua, Z., Mali, S., McLoughlin, F. & Vierstra, R. D. ATG8-binding UIM proteins define a new class of autophagy adaptors

and receptors. _Cell_ 177, 766–781 (2019). Article  CAS  Google Scholar  * Pantazopoulou, M., Boban, M., Foisner, R. & Ljungdahl, P. O. Cdc48 and Ubx1 participate in a pathway associated

with the inner nuclear membrane that governs Asi1 degradation. _J. Cell Sci._ 129, 3770–3780 (2016). Article  CAS  Google Scholar  * Berner, N., Reutter, K. R. & Wolf, D. H. Protein

quality control of the endoplasmic reticulum and ubiquitin-proteasome-triggered degradation of aberrant proteins: yeast pioneers the path. _Annu. Rev. Biochem._ 87, 751–782 (2018). Article 

CAS  Google Scholar  * Liu, Y. & Li, J. Endoplasmic reticulum-mediated protein quality control in _Arabidopsis_. _Front. Plant Sci._ 5, 162 (2014). ADS  PubMed  PubMed Central  Google

Scholar  * Meusser, B., Hirsch, C., Jarosch, E. & Sommer, T. ERAD: the long road to destruction. _Nat. Cell Biol._ 7, 766–772 (2005). Article  CAS  Google Scholar  * Ling, Q. et al.

Ubiquitin-dependent chloroplast-associated protein degradation in plants. _Science_ https://doi.org/10.1126/science.aav4467 (2019). * Park, S., Rancour, D. M. & Bednarek, S. Y. In planta

analysis of the cell cycle-dependent localization of AtCDC48A and its critical roles in cell division, expansion, and differentiation. _Plant Physiol._ 148, 246–258 (2008). Article  CAS 

Google Scholar  * de Leeuw, R., Gruenbaum, Y. & Medalia, O. Nuclear lamins: thin filaments with major functions. _Trends Cell Biol._ 28, 34–45 (2018). Article  Google Scholar  * Gu, Y.

& Innes, R. W. The keep on going protein of _Arabidopsis_ recruits the enhanced disease resistance1 protein to trans-Golgi network/early endosome vesicles. _Plant Physiol._ 155,

1827–1838 (2011). Article  CAS  Google Scholar  * Gu, Y. & Innes, R. W. The keep on going protein of _Arabidopsis_ regulates intracellular protein trafficking and is degraded during

fungal infection. _Plant Cell_ 24, 4717–4730 (2012). Article  CAS  Google Scholar  * Gu, Y. et al. Nuclear pore permeabilization is a convergent signaling event in effector-triggered

immunity. _Cell_ 166, 1526–1538 (2016). Article  CAS  Google Scholar  Download references ACKNOWLEDGEMENTS We thank Dr. Sheila McCormick for critical reading of the manuscript and Dr. Qi-Jun

Chen for providing the pCBC-DT1T2 and pHEE401 vector. This research was supported by grants from Innovative Genomics Institute at the University of California Berkeley and Tsinghua-Peking

Joint Center for Life Sciences at Tsinghua University. AUTHOR INFORMATION AUTHORS AND AFFILIATIONS * Tsinghua-Peking Joint Center for Life Sciences, Center for Plant Biology, School of Life

Sciences, Tsinghua University, Beijing, China Aobo Huang, Xuetao Shi, Jinheng Zhu, Xiaohan Yan & Huiqin Chen * Department of Plant and Microbial Biology, University of California,

Berkeley, CA, USA Yu Tang, Min Jia & Yangnan Gu * Innovative Genomics Institute, University of California, Berkeley, CA, USA Yangnan Gu Authors * Aobo Huang View author publications You

can also search for this author inPubMed Google Scholar * Yu Tang View author publications You can also search for this author inPubMed Google Scholar * Xuetao Shi View author publications

You can also search for this author inPubMed Google Scholar * Min Jia View author publications You can also search for this author inPubMed Google Scholar * Jinheng Zhu View author

publications You can also search for this author inPubMed Google Scholar * Xiaohan Yan View author publications You can also search for this author inPubMed Google Scholar * Huiqin Chen View

author publications You can also search for this author inPubMed Google Scholar * Yangnan Gu View author publications You can also search for this author inPubMed Google Scholar

CONTRIBUTIONS A.H. and Y.G. designed the research. A.H., Y.T., X.S., M.J., J.Z., X.Y., and H.C. performed the experiments. A.H. and Y.G. analyzed the data and wrote the paper. CORRESPONDING

AUTHOR Correspondence to Yangnan Gu. ETHICS DECLARATIONS COMPETING INTERESTS The authors declare no competing interests. ADDITIONAL INFORMATION PEER REVIEW INFORMATION _Nature

Communications_ thank the anonymous reviewers for their contribution to the peer review of this work. Peer review reports are available. PUBLISHER’S NOTE Springer Nature remains neutral with

regard to jurisdictional claims in published maps and institutional affiliations. SUPPLEMENTARY INFORMATION PEER REVIEW FILE SUPPLEMENTARY INFORMATION DESCRIPTION OF ADDITIONAL

SUPPLEMENTARY FILES SUPPLEMENTARY DATA 1 SUPPLEMENTARY DATA 2 SUPPLEMENTARY DATA 3 REPORTING SUMMARY SOURCE DATA SOURCE DATA RIGHTS AND PERMISSIONS OPEN ACCESS This article is licensed under

a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate

credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article

are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and

your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this

license, visit http://creativecommons.org/licenses/by/4.0/. Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE Huang, A., Tang, Y., Shi, X. _et al._ Proximity labeling proteomics

reveals critical regulators for inner nuclear membrane protein degradation in plants. _Nat Commun_ 11, 3284 (2020). https://doi.org/10.1038/s41467-020-16744-1 Download citation * Received:

10 February 2020 * Accepted: 20 May 2020 * Published: 29 June 2020 * DOI: https://doi.org/10.1038/s41467-020-16744-1 SHARE THIS ARTICLE Anyone you share the following link with will be able

to read this content: Get shareable link Sorry, a shareable link is not currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt content-sharing

initiative