Nt-acetylation-independent turnover of SQUALENE EPOXIDASE 1 by Arabidopsis DOA10-like E3 ligases

Abstract The acetylation-dependent (Ac/)N-degron pathway degrades proteins through recognition of their acetylated N-termini (Nt) by E3 ligases called Ac/N-recognins. To date, specific Ac/N-recognins have not been defined in plants. Here we used molecular, genetic, and multiomics approaches to characterize potential roles for Arabidopsis (Arabidopsis thaliana) DEGRADATION OF ALPHA2 10 (DOA10)-like E3 ligases in the Nt-acetylation-(NTA)-dependent turnover of proteins at global- and protein-specific scales. Arabidopsis has two endoplasmic reticulum (ER)-localized DOA10-like proteins. AtDOA10A, but not the Brassicaceae-specific AtDOA10B, can compensate for loss of yeast (Saccharomyces cerevisiae) ScDOA10 function. Transcriptome and Nt-acetylome profiling of an Atdoa10a/b RNAi mutant revealed no obvious differences in the global NTA profile compared to wild type, suggesting that AtDOA10s do not regulate the bulk turnover of NTA substrates. Using protein steady-state and cycloheximide-chase degradation assays in yeast and Arabidopsis, we showed that turnover of ER-localized SQUALENE EPOXIDASE 1 (AtSQE1), a critical sterol biosynthesis enzyme, is mediated by AtDOA10s. Degradation of AtSQE1 in planta did not depend on NTA, but Nt-acetyltransferases indirectly impacted its turnover in yeast, indicating kingdom-specific differences in NTA and cellular proteostasis. Our work suggests that, in contrast to yeast and mammals, targeting of Nt-acetylated proteins is not a major function of DOA10-like E3 ligases in Arabidopsis and provides further insight into plant ERAD and the conservation of regulatory mechanisms controlling sterol biosynthesis in eukaryotes.

The addition of an acetyl group has the effect of neutralizing N-terminal charge and increasing hydrophobicity, which can influence protein fate in several ways, for example, by promoting protein-protein interactions and directing protein localization by increasing membrane affinity (Linster and Wirtz 2018;Ree et al. 2018).NTA also impacts protein folding, with deletions of NATA and particularly NATB causing the accumulation of misfolded protein aggregates in yeast (Friedrich et al. 2021).Furthermore, NTA has been shown to promote or prevent protein degradation depending on the protein target and cellular context.
In yeast and humans, acetylation of N-termini can destabilize certain proteins through the creation of Ac/N-degrons that target them for proteolysis via the acetylation-dependent (Ac/) N-degron pathway (Hwang et al. 2010;Shemorry et al. 2013;Park et al. 2015).This degradation pathway was identified in yeast (Saccharomyces cerevisiae), where DEGRADATION OF ALPHA2 10 (DOA10) and NEGATIVE ON TATA-LESS 4 (NOT4) E3 ligases were shown to function as Ac/N-recognins that target substrates via recognition of their Nt-acetylated N-termini (Hwang et al. 2010;Shemorry et al. 2013).Ac/ N-degron pathway substrates have also been identified in humans (Homo sapiens) and are recognized by the human ScDOA10 homolog, MEMBRANE-ASSOCIATED RING-CH-TYPE FINGER 6 (MARCH6)/TEB4 (Park et al. 2015;Nguyen et al. 2019).In Arabidopsis (Arabidopsis thaliana), NTA of an MMD-initiating isoform of SUPPRESSOR OF NPR1, CONSTITUTIVE 1 (SNC1) by NATA was shown to induce degradation, suggesting the Ac/N-degron pathway may also function in plants, though to date no plant Ac/N-recognins have been identified (Xu et al. 2015).Interestingly however, NTA of an alternative MD-initiating isoform of SNC1 by NATB was shown to have a stabilizing effect, indicating that Nt-variants of the same protein can be differentially targeted for degradation (Gibbs 2015;Xu et al. 2015).Recently, the NATA-interacting HUNTINGTIN INTERACTING PROTEIN K (HYPK) protein in rice (Oryza sativa) was also shown to be degraded following N-terminal acetylation (Gong et al. 2022).Since most cellular proteins are Nt-acetylated, degradation via the Ac/N-degron pathway is proposed to be conditional, with substrates only degraded when their acetylated N-termini are not internalized within a protein's structure or shielded by a binding partner (Shemorry et al. 2013).
The discovery of the Ac/N-degron pathway partially conflicted with the historical view that NTA increases protein halflife by blocking ubiquitination of the N-terminus (Hershko et al. 1984).Recent studies have also suggested that NTA does not act as a broad or universal degradation signal.High-throughput screening studies have independently shown that unstructured NTA reporter substrates of NATA and NATB were generally stable and that mutations of NATA or NATB did not increase the abundance of their endogenous targets (Kats et al. 2018;Friedrich et al. 2021).Indeed, NTA has also been reported to block other Nt-processing events such as Met-excision, arginylation, and the binding of N-recognins, thereby potentially protecting proteins from other degradative mechanisms such as the Arg/N-degron pathway (Kim et al. 2014;Park et al. 2015;Kats et al. 2018).One such protein is Arabidopsis SIGMA FACTOR-BINDING PROTEIN1 (SIB1), which is stabilized following NTA by NATB (Li et al. 2020).Kats et al. (2018) also observed that while mutation of ScDOA10 did stabilize many normally unstable reporter proteins, turnover of these peptides was largely defined by Nt-hydrophobicity rather than NTA itself.Additionally, in human cells, NTA by NATA was shown to protect nascent proteins from degradation by preventing their unwanted interaction with IAP E3 ligases that might otherwise trigger ectopic apoptosis (Mueller et al. 2021).Moreover, an analogous proteome-wide role for NTA in protein stabilization was also recently uncovered in Arabidopsis, where stress-responsive NATA activity (Linster et al. 2015) was shown to mask non-Ac/N-degrons that would otherwise target NATA substrates for proteasomal degradation by as-yet-unknown E3 ligases (Gibbs et al. 2022;Linster et al. 2022).
The best characterized Ac/N-recognin, ScDOA10, is a RING-type E3 ligase that localizes to the endoplasmic reticulum (ER) and nuclear envelope and which was identified as a major component of the endoplasmic reticulum-associated protein degradation (ERAD) system that degrades misfolded ER proteins (Swanson et al. 2001).ScDOA10 is one of two ERAD E3 ligases and is primarily responsible for the ubiquitination of proteins with cytosolic degrons (ERAD-C) (Hirsch et al. 2009;Strasser 2018) although targeting of degrons within the ER membrane and retrotranslocase activity have also been reported for ScDOA10 (Habeck et al. 2015;Schmidt et al. 2020).Two putative homologs of ScDOA10 have been identified in Arabidopsis: AtDOA10A, also known as ECERIFERUM9 (CER9)/SUPPRESSOR OF DRY2 DEFECTS1 (SUD1), and AtDOA10B (Liu et al. 2011).Atdoa10a mutants display a range of phenotypes, including altered cuticular wax composition, improved drought tolerance, and ABA hypersensitivity during germination (Lü et al. 2012;Zhao et al. 2014).Mutations in AtDOA10A were also shown to repress the pleiotropic phenotypes caused by a point mutation in the sterol biosynthesis gene SQUALENE EPOXIDASE 1 (SQE1)/DROUGHT HYPERSENSITIVE2 (DRY2) by downregulating 3-HYDROY-3-METHYLGLUTARYL-COA REDUCTASE (HMGR), an upstream enzyme of the pathway (Doblas et al. 2013).It is still unclear if either AtDOA10 homolog plays a major role in the plant ERAD system (Li et al. 2017;Huber et al. 2022).Furthermore, potential Ac/N-recognin functions for AtDOA10s have not yet been investigated, and no physiological substrates of AtDOA10s have been identified.
Here, we sought to functionally characterize both putative AtDOA10 orthologs with a view to establishing whether they function as Ac/N-recognin E3 ligases of the as-yet-uncharacterized Ac/N-degron pathway in plants.
Our global-and protein-specific results uncover a previously unknown function for AtDOA10s in the homeostatic regulation of sterol biosynthesis through controlling AtSQE1 turnover and suggest that their primary function in Arabidopsis is unrelated to the Ac/N-degron pathway and the bulk degradation of Nt-acetylated proteins.

Structure, functional conservation, and phylogeny of Arabidopsis DOA10-like proteins
To identify putative DOA10 homologs in Arabidopsis, we searched for protein sequences with homology to full-length ScDOA10 from Saccharomyces cerevisiae, which, in accordance with previous studies, identified two DOA10-like proteins encoded by the genes At4g34100 (AtDOA10A/CER9/ SUD1) and At4g32670 (AtDOA10B) (Liu et al. 2011;Lü et al. 2012;Doblas et al. 2013).These share 22.87% and 19.23% amino acid identity with ScDOA10, respectively, and 35.66% with each other (Fig. 1, A and B and Supplemental Fig. S1).Identity is particularly high in the N-terminal RING-CH domain, an atypical variant of the classic RING domain that provides ubiquitin ligase activity, and the TEB4/DOA10 (TD) domain, which influences interactions with the cognate ubiquitin conjugase 6 (UBC6) E2 enzyme in yeast (Fig. 1B) (Kreft and Hochstrasser 2011;Doblas et al. 2013).Both AtDOA10s also have an extended C-terminal region containing 13 to 16 predicted transmembrane (TM) domains (Supplemental Fig. S2, A and B), similar to the experimentally confirmed 14 TM domains in ScDOA10 (Fig. 1A) (Kreft et al. 2006).
To determine whether AtDOA10A and AtDOA10B represent functional homologs of ScDOA10, we assessed their capacity to complement the yeast Scdoa10Δ mutant.We cloned both proteins as C-terminal GFP fusions driven by the GPD promoter, transformed them into Scdoa10Δ, and confirmed expression using RT-PCR (Fig. 1C).In growth assays, Scdoa10Δ displayed relative insensitivity to hygromycin, which was reverted in mutants expressing AtDOA10A-GFP (to a greater extent than wild-type [WT] yeast, possibly due to overexpression of the transgene), but unaffected in lines expressing AtDOA10B-GFP (Fig. 1D).Thus, AtDOA10A, but not the C-terminally truncated AtDOA10B (Fig. 1A), is able to compensate for the loss of endogenous ScDOA10 activity in yeast, indicating at least partial conservation of function for this putative ortholog.
To understand the nature of the Arabidopsis DOA10 gene duplication, we constructed a phylogenetic tree of DOA10like protein sequences identified via BLASTP from a range of diploid flowering plant genomes, including diverse monocots, dicots, and several members of the Brassicaceae family (Fig. 1E).We found that many, though not all, plant species had two DOA10-like proteins.While most DOA10-like sequences clustered into defined branches that split the monocots and dicots, AtDOA10B was part of a separate Brassicaceaespecific clade, grouping with similar DOA10B-like sequences from lyre-leaved rock cress (Arabidopsis lyrata), field mustard (Brassica rapa), and pink shepherd's purse (Capsella rubella).
In contrast, where two putative orthologs were identified in other species, both sequences occurred together in speciesspecific pairs within the main AtDOA10A-like clade-e.g.pineapple (Ananas comosus), poplar (Populus trichocarpa), and soybean (Glycine max).This suggests that the truncated DOA10B-like variant emerged in the Brassicaceae lineage.

AtDOA10A and AtDOA10B are broadly expressed and localize to the endoplasmic reticulum
We developed transgenic lines expressing C-terminal GUS fusions of AtDOA10s driven by 2 kb of the endogenous promoter (pAtDOA10A/B::AtDOA10A/B-GUS).Histochemical staining revealed that both proteins are broadly detectable in 7-and 14-day-old seedlings, particularly in roots, which showed enrichment in the primary root meristem, lateral root primordia, and vasculature (Fig. 2A).A complementary reverse transcription quantitative PCR (RT-qPCR) analysis of relative AtDOA10A/B mRNA abundance also identified transcripts for both proteins across a range of seedling and adult tissues, suggesting broad roles for AtDOA10s in diverse cell types (Fig. 2B).Corroborating the reduced levels of AtDOA10B-GUS staining relative to AtDOA10A-GUS, AtDOA10B mRNA abundance was much lower than AtDOA10A (approx.20-fold) across all tissue types.

Generation and phenotypic assessment of AtDOA10A and B mutants
We obtained homozygous Atdoa10a and Atdoa10b T-DNA insertional mutants and confirmed the knockouts by RT-PCR (Fig. 3A and Supplemental Fig. S3A).Although only AtDOA10A was able to complement Scdoa10Δ, we postulated that both AtDOA10s could have overlapping or redundant functions in Arabidopsis.As such, we attempted to make a double mutant, but the proximity of the two genes (separated by 0.57 Mb) meant that a crossover event was likely to be very rare; in accordance with this, we were unable to identify any double mutants and instead took an RNAi approach.We designed two different RNAi constructs that targeted the first and second exons of AtDOA10B (Fig. 3A) and transformed these into Atdoa10a.RT-qPCR analysis of the derived progeny led to the identification of two independent lines with reduced AtDOA10B mRNA (∼50% Col-0 levels): Atdoa10a/b RNAi 3-7 and RNAi 4-2 (Fig. 3B).Despite screening several lines, we did not identify any with stronger depletion.
No major phenotypic differences between lines were observed when grown under standard conditions, though mutant rosettes were slightly smaller than WT rosettes, in accordance with previous observations for Atdoa10a (Fig. 3C) (Huber et al. 2022).AtDOA10A was previously linked to ABA signaling and the control of cuticle development, with Atdoa10a mutants displaying seed ABA hypersensitivity and drought tolerance phenotypes (Lü et  2012; Zhao et al. 2014).We also observed ABA hypersensitivity of Atdoa10a seeds (Fig. 3D) and showed that WT sensitivity was restored in Atdoa10a lines complemented with pDOA10A:: AtDOA10A-YFP (Supplemental Fig. S3B).In contrast, the Atdoa10b single mutant had no obvious phenotype, and ABA sensitivity of Atdoa10a/b RNAi 4-2 resembled that of the Atdoa10a single mutant, suggesting that AtDOA10A and AtDOA10B do not have additive or redundant roles in ABA-related responses.

RNA-sequencing and Nt-acetylome profiling indicate that AtDOA10s do not regulate global turnover of Nt-acetylated proteins
We carried out an RNA-seq analysis of 10-day-old Atdoa10a/ b RNAi 4-2 seedlings grown under long-day conditions, which identified 447 differentially expressed genes (DEGs; 217 up, 230 down) relative to Col-0 (False Discovery Rate [FDR] adjusted P-value [q] of <0.05) (Fig. 4A and Supplemental Data Set S1).Among these, 89 had a fold change of 2 or more.This modest number of DEGs likely reflects the fact that plants were grown under ambient, non-stressed conditions.Nonetheless, Gene Ontology (GO) analysis uncovered several enriched GO terms that are consistent with potential roles for AtDOA10s in protein quality control or ERAD, including cellular component term perinuclear region of the cytoplasm (18.1-fold enrichment, 2.05E-07 FDR), biological process terms protein folding (4.63-fold enrichment, 6.31E-03 FDR) and protein refolding (10.18-fold enrichment, 1.50E-02 FDR), and molecular function term unfolded protein binding (5.14-fold enrichment, 1.61E-02 FDR).We also carried out RNA-seq analysis on 10-day-old NAT-depleted plants: amiNAA10 (i.e.NATA) knockdown and Atnaa20 (i.e.NATB) mutant seedlings.We hypothesized that there might be some overlap in the transcriptome of Atdoa10a/b and nat mutants, due to shared ectopic stabilization of Ac/ N-degron protein targets.As expected, given NATA's role in acetylating 40% of cytosolic proteins, amiNAA10 seedlings had the greatest number of DEGs (7,139), with Atnaa20 having fewer (2,486).Interestingly, nearly 70% of the Atdoa10a/b RNAi-annotated DEGs (307/444) were also differentially expressed in either NAT mutant transcriptome (Fig. 4B), with 75% of Atdoa10a/b-amiNAA10 and 78% of Atdoa10a/ b-naa20 shared DEGs occurring in the same direction (i.e. up in both or down in both; Supplemental Data Set S1).
If AtDOA10s function as general Ac/N-recognins of the Ac/ N-degron pathway, an accumulation of Nt-acetylated proteins would be expected in plants lacking AtDOA10 function compared to Col-0.This could manifest as either an increase in the total levels of Nt-acetylated proteins or increased ratio of acetylated to non-acetylated variants of a particular protein(s).We performed quantitative Nt-acetylome profiling on total protein extracts from both Col-0 and doa10a/b RNAi 4-2 seedlings using the "stable isotope labeling protein N-terminal acetylation quantification (SILProNAQ)" method (Bienvenut et al. 2017a), with data processed using the EnCOUNTer tool (Bienvenut et al. 2017b) (Supplemental Data Set S2).A total of 651 and 594 N-termini were identified in Col-0 and Atdoa10a/b RNAi 4-2, respectively, with 426 common to both lines (Fig. 4, C and D).In both genotypes, approximately half of these N-termini were Nt-acetylated (48% in Col-0 and 49% in Atdoa10a/b).We were able to quantify a total of 342 unique N-termini (277 in Col-0 and 256 in Atdoa10a/b RNAi 4-2).In Col-0 58% (161/277) were either fully or partially acetylated, which was similar to 59% observed for Atdoa10a/b RNAi 4-2 (151/256).Global NTA variation comparisons based on either N-terminal position (Fig. 4E) or cellular sub-compartment (Supplemental Fig. S4) showed no differences in overall NTA level in Atdoa10a/b RNAi 4-2 vs Col-0, and a similar distribution in NTA-frequency was observed for all natural amino acids at the N-terminal position (Fig. 4F).Moreover, relative quantification of Nt-acetylated peptides shared between Atdoa10a/b RNAi 4-2 and WT identified no proteins with significantly (i.e.FDR < 5%) increased or decreased NTA (Fig. 4G).Collectively, these Nt-profiling findings suggest that strong depletion of AtDOA10 levels does not affect the overall turnover of Nt-acetylated proteins.

Proteolytic turnover of AtSQE1 requires AtDOA10A and AtNAA20 in heterologous yeast degradation assays
While the Atdoa10a/b RNAi 4-2 double mutant had no clear differences in global NTA, AtDOA10s could still play a role in targeting specific Nt-acetylated protein substrates for degradation.We took a targeted approach to identify a potential physiological substrate to use for investigating protein turnover in finer molecular detail.Mutations in AtDOA10A were previously shown to suppress phenotypic defects in the dry2 mutant, which lacks SQE1 activity.Here, defects arise due to a build-up of toxic intermediates, and a secondary mutation in AtDOA10A alleviates this by downregulating HMGR enzyme activity, which functions several steps ahead of squalene synthesis (Doblas et al. 2013).Moreover, in yeast and mammals, SQE homologs (Ergosterol Biosynthesis 1 [ERG1] in yeast and squalene monooxygenase [SM] in mammals) localize to the ER membrane and are direct proteolytic targets of ScDOA10 and HsMARCH6/TEB4, respectively (Foresti et al. 2013;Zelcer et al. 2014;Sharpe et al. 2020).Similarly, Arabidopsis SQE1 is predicted to have transmembrane regions and localize to the ER, although it does not contain a predicted N-terminal secretory signal peptide (Supplemental Fig. S2, C and D).Given the evolutionary conservation of DOA10-like E3 ligases and their roles in modulating sterol biosynthetic pathways, we postulated that Arabidopsis SQE1 turnover might also be regulated by AtDOA10s.
We used WT and mutant yeast strains as a heterologous system for expressing AtSQE1 and monitoring its stability by immunoblotting.Steady-state levels of AtSQE1-HA were higher in Scdoa10Δ than WT yeast, and this could be reverted when AtSQE1-HA was co-expressed with AtDOA10A-YFP, but not AtDOA10B-YFP (Fig. 5, A and B).Next, we monitored AtSQE1-HA turnover rates using cycloheximide (CHX) chase assays, which revealed rapid degradation of AtSQE1-HA in WT yeast cells, but relative stabilization in Scdoa10Δ (Fig. 5C).Moreover, the enhanced stability was partially reverted when AtDOA10A-YFP was co-transformed into Scdoa10Δ (Fig. 5D).
AtSQE1 initiates with the residues Met-Glu-(ME-) and should retain its initiator Met during translation and be targeted by ribosome-associated NATB (comprising NAA20 catalytic and NAA25 auxiliary subunits) (Aksnes et al. 2019).It was previously shown that 99% of ME-initiating proteins undergo NTA in humans (Aksnes et al. 2016), with similarly high numbers in yeast and Arabidopsis (Supplemental Fig. S5 and Supplemental Data Set S3).It was also shown that human HsNAA20 can complement Arabidopsis Atnaa20 mutants but that yeast ScNAA20 cannot (Huber et al. 2020).Whether AtNAA20 can compensate for loss of ScNAA20 function is unknown.To investigate whether AtSQE1-HA turnover is mediated by Nt-acetylation and the Ac/N-degron pathway, we also monitored AtSQE1-HA stability in the yeast Scnaa20Δ mutant (Fig. 5, C and E).Similar to Scdoa10Δ, and potentially consistent with the Ac/N-degron pathway, we saw strong enhancement of AtSQE1-HA stability relative to WT, which was almost completely reverted when AtNAA20 was co-expressed.This indicates that NATB activity is required for AtSQE1 turnover and reveals that Arabidopsis AtNAA20 can functionally replace ScNAA20.

Impact of N-terminal mutagenesis on AtSQE1 stability suggests indirect effects of Nt-acetyltransferases on protein turnover in yeast
To further investigate the connection between NTA and AtSQE1 degradation, we developed a series of Nt-mutagenized variants of AtSQE1 that are predicted to be blocked for NTA or which are targeted by different cognate NATs (Fig. 6A, Supplemental Fig. S5, and Supplemental Data Set S3): (1) MP-SQE1, which should undergo Met-excision by methionine aminopeptidases (MetAPs) but no further Nt-acetylation, as Nt-Proline residues are not acetylated (Goetze et al. 2009); (2) MK-SQE1, which should be rarely acetylated (Arnesen et al. 2009); and (3) MA-SQE1, which would instead be Nt-acetylated by NATA/NAA10 following initiator Met removal by MetAP.We hypothesized that MP-and MK-SQE1-HA would be stable in WT yeast if turnover is dependent on NTA but instead found that their steady-state levels were in fact reduced relative to WT ME-AtSQE1-HA (Fig. 6B), perhaps due to codon differences impacting translation (Kozak 1997).Moreover, when expressed in Scdoa10Δ, relative abundance increased for all proteins, but the ratios between them were maintained, suggesting that ScDOA10 targets all three Nt-variants and that their direct NTA is not critical for degradation.
In accordance with these steady-state analyses (Fig. 6B), we found that mutant MP-AtSQE1-HA rapidly degraded following CHX treatment, similar to WT ME-AtSQE1-HA (Fig. 6C).We also observed stabilization of WT ME-AtSQE1-HA in Scnaa10Δ mutant cells (Fig. 6D), which lack a functional NATA enzymatic complex; this was unexpected, since ME-is not a target sequence for NATA activity.Finally, we examined the stability of mutant MA-AtSQE1-HA, where the N-terminus is remodeled from a NATB to a NATA target.Like all other variants, MA-AtSQE1-HA was unstable in WT yeast but was surprisingly still stable in the non-cognate Scnaa20Δ mutant in addition to Scdoa10Δ (Fig. 6, E  and F).Collectively, these assays indicate that degradation of AtSQE1 via DOA10 in yeast is indirectly influenced by NATA and NATB activity.

Nt-acetylation-independent turnover of AtSQE1 by AtDOA10 in Arabidopsis
To test AtSQE1 stability in planta, we generated a range of stable transgenic Arabidopsis lines expressing Nt-variants of AtSQE1-Myc driven by the constitutive 35S CaMV promoter.WT ME-AtSQE1-Myc was expressed in Col-0, Atdoa10a, Atdoa10b, Atdoa10a/b RNAi 4-2, and Atnaa20, while mutant non-acetylatable MP-AtSQE1-Myc was expressed in Col-0 only.For each construct we identified two independent transgenics and confirmed their expression by RT-qPCR (Supplemental Fig. S6A).Both the ME-and MP-Nt-variants of AtSQE1-Myc localized to the ER in Col-0 (Fig. 7A).Thus, AtSQE1 resides in the same cellular compartment as AtDOA10s (like in yeast and mammals), and the introduced E2P N-terminal mutation does not disrupt this subcellular targeting.CHX chase assays revealed that WT ME-AtSQE1-Myc was unstable after 24 h of translational shutdown with CHX in all genetic backgrounds tested, except for the Atdoa10a/b RNAi 4-2 double mutant (Fig. 7B).This was consistent across two independent transgenic lines (Fig. 7B and Supplemental Fig. S6B).A shorter CHX experiment corroborated this finding, showing that WT ME-AtSQE1-Myc is turned over in WT within 6 h, but remains stable in the Atdoa10a/b RNAi 4-2, or if coincubated with proteasome inhibitor bortezomib (Fig. 7C and Supplemental Fig. S6C).Mutant MP-AtSQE1-Myc was also unstable in Col-0 (Fig. 7B).These data reveal that proteasomal degradation of AtSQE1 requires both AtDOA10 proteins in planta and that this turnover is neither dependent on indirect NAT activity (in contrast to yeast; Figs. 5 and 6) nor direct NTA of AtSQE1 by NATB.
Potentially consistent with the fact that AtSQE1 accumulates in Atdoa10a/b RNAi 4-2, we observed extreme epinasty in seedlings of this mutant relative to Col-0, a phenotype that has previously been shown to coincide with altered sterol biosynthesis (Fig. 7D) (Carland et al. 2010).Precise control of sterol synthesis pathway enzymes is required to regulate and maintain appropriate levels of phytosterols and their intermediates in Arabidopsis (De Vriese et al. 2021).In human and yeast systems, SQE expression and stability are influenced by the accumulation or consumption of different sterols and pathway intermediates, but how such feedback works in Arabidopsis is currently unknown (Leber et al. 2001;Gill et al. 2011;Foresti et al. 2013;Scott et al. 2020).We investigated whether chemical inhibition of Arabidopsis sterol synthesis impacts AtSQE1 stability via AtDOA10s through treating seedlings with the chemical lauryldimethylamine oxide (LDAO), a dual inhibitor of both the cycloartenol synthase (CAS) enzymes, which cyclize 2,3-oxidosqualene (the direct product of AtSQE1) to produce cycloartenol, and cycloeucalenol cycloisomerase (CPI) enzymes, which act further downstream (Darnet et al. 2020).We observed extremely rapid turnover of ME-AtSQE1-Myc protein in response to LDAO treatment, occurring within just 30 min (Fig. 7E).This was specific to ME-AtSQE1-Myc and not a consequence of general protein turnover nor ER disruption due to LDAO's detergent properties, since ER-localized CNX1/2 proteins were unaffected.Interestingly however, LDAO was still able to induce rapid turnover of ME-AtSQE1-Myc in Atdoa10a/b RNAi 4-2 (Fig. 7E).We also directly inhibited AtSQE1 using terbinafine, a non-competitive inhibitor of squalene epoxidase enzymes (Nowosielski et al. 2011); this led to strong accumulation of AtSQE1-Myc (Fig. 7F), likely caused by positive feedback through reduced degradation given that expression was driven by the constitutive 35S CaMV promoter.

Discussion
In yeast and humans, DOA10 E3 ligases function as Ac/ N-recognins that target functionally diverse proteins for degradation via recognition of their acetylated N-termini (Hwang et al. 2010;Shemorry et al. 2013;Park et al. 2015).In plants, the Ac/N-degron pathway and its associated Ac/N-recognins are yet to be characterized, although it was previously shown that NTA of rice HYPK and a specific Nt-isoform of Arabidopsis SNC1 triggers their degradation (Xu et al. 2015;Gong et al. 2022).While HYPK and SNC1 are potential plant Ac/N-degron pathway targets, the Arabidopsis SIB1 protein was recently shown to be stabilized by NATB-mediated NTA (Li et al. 2020).Moreover, large-scale studies in yeast, humans, and plants have established broader roles for NTA in proteome stabilization (Kats et al. 2018;Friedrich et al. 2021;Mueller et al. 2021;Gibbs et al. 2022;Linster et al. 2022).Therefore, the relationship between protein NTA and degradation is complex and is likely to vary depending on protein identity and cellular context.Here, we investigated Arabidopsis AtDOA10-like E3 ligases, focusing specifically on their potential function as Ac/ N-recognins.Nt-acetylome profiling revealed no apparent differences in the accumulation of Nt-acetylated proteins in Atdoa10a/b RNAi vs Col-0, suggesting that depletion of AtDOA10 proteins does not influence the bulk turnover of Nt-acetylated proteins.However, the proteins identified and quantified in our comparative Nt-acetylome analysis represent only a small proportion of total potential Ac/N-degron substrates, which means we cannot rule out that AtDOA10s may have specific roles in targeting a more constrained set of Nt-acetylated proteins.Alternatively, other E3 ligases, for example, putative orthologs of the NOT4 Ac/N-recognins (Shemorry et al. 2013;Gibbs et al. 2016), may function as primary Ac/N-recognins in plants.
We identified the ER-resident AtSQE1 protein, a ratelimiting enzyme in the sterol biosynthesis pathway (Rasbery et al. 2007), as a target of AtDOA10s.Turnover of AtSQE1 requires functional AtDOA10s and was not linked to acetylation of its N-terminus in planta, since its stability was unaltered when its putative cognate NAT, NATB, was deleted (Fig. 7B).Surprisingly however, we found that AtSQE1 stability was indirectly influenced by NAT activity when heterologously expressed in yeast.Partial or complete NATA or NATB inactivation led to enhanced stabilization regardless of the Nt-variant being assayed (Figs. 5 and 6), which may be due to indirect effects on other proteins linked to translational or proteolytic machineries; the ERAD protein ScDer1, for example, requires NTA by NATB for stability (Zattas et al. 2013).Our cross-kingdom analysis of AtSQE1 stability has therefore uncovered differences in the ways in which NTA can influence proteostasis in yeast and plants, suggesting that caution should be applied when investigating the connection between NTA and protein turnover and particularly when extrapolating between organisms.
In contrast to yeast and humans, the Arabidopsis genome encodes for two DOA10-like E3 ligases.AtDOA10A, but not AtDOA10B, was able to complement Scdoa10Δ yeast, with respect to both hygromycin sensitivity and turnover of heterologously expressed AtSQE1 (Figs. 1 and 5).While both AtDOA10s are smaller than ScDOA10, AtDOA10B is particularly truncated (Fig. 1A) and appears to be restricted to the Brassicaceae clade (Fig. 1E), which could explain its inability to complement Scdoa10Δ.This may be the result of incompatibility between AtDOA10B and components of the endogenous yeast ubiquitination machinery, since in Arabidopsis AtSQE1 degradation was only inhibited in the Atdoa10a/b RNAi double mutant.Despite this example of functional redundancy, other evidence points to additional paralog-specific activities.For instance, Atdoa10a single mutants display a range of phenotypes that do not manifest in Atdoa10b and which are not amplified in Atdoa10a/b RNAi lines, including altered cuticular wax composition and strong ABA hypersensitivity (Fig. 3D) (Lü et al. 2012;Zhao et al. 2014).DOA10s in yeast and humans are major E3 ligases of the ERAD system (Ravid et al. 2006), while roles for AtDOA10s in the Arabidopsis ERAD system are still unclear (Liu et al. 2011;Li et al. 2017;Huber et al. 2022).This may be due to functional redundancy, a concept supported by our observations that turnover of ER-resident AtSQE1 is dependent on both AtDOA10s.Interestingly though, AtDOA10B was shown to physically associate with the ERAD-associated ubiquitin conjugase 32 (UBC32) (Cui et al. 2012) and be transcriptionally induced by L-azetidine-2-carboxylic acid (AZC), a proline analogue that causes protein misfolding (Kim et al. 2017).We also observed significant upregulation of AtDOA10B, but not AtDOA10A, in response to the ERAD elicitor tunicamycin (Supplemental Fig. S7).This suggests that AtDOA10s have both redundant and distinct roles linked to different cellular processes and that the AtDOA10B paralog may have evolved to take on a more prominent role in stress-associated, rather than constitutive, ERAD in Arabidopsis.Further analysis of the transcriptomic, proteomic, and physiological response of Atdoa10a and b mutants to ER and protein misfolding stresses will shed further light on the roles these proteins play in ERAD and protein homeostasis.
Sterol biosynthesis is sensitive to fluctuations in enzyme activity and substrate availability at each stage.For example, a build-up of cholesterol in animals feeds back to downregulate the pathway and redirect precursor flux, while increases in lanosterol abundance has a similar effect in yeast (Gill et al. 2011;Foresti et al. 2013;Scott et al. 2020).In plants, a build-up of squalene and/or its precursors is toxic but can be counteracted by a reduction of HMGR enzyme activity, which is enhanced when AtDOA10A is knocked out (Doblas et al. 2013).Our LDAO assays reveal that ectopic build-up of downstream sterol intermediates also triggers feedback mechanisms to downregulate sterol production in plants, in this case promoting AtSQE1 destabilization, possibly through an alternative E3 ligase or an autophagic mechanism (Fig. 7E).In contrast, direct inhibition of AtSQE1 with terbinafine led to increased accumulation of AtSQE1 (Fig. 7F).Interestingly, ScDOA10 was previously shown to target multiple enzymes of the sterol biosynthesis pathway in yeast, indicating that it is a master regulator of this biochemical pathway (Scott et al. 2020).The fact that AtDOA10s influence both HMGR activity and SQE1 stability also points to central functions for DOA10s in coordinating sterol production at several steps in plants (Fig. 8).Whether AtDOA10s control turnover of other Arabidopsis SQEs (Rasbery et al. 2007) remains to be determined, though the Brassicaceae-specific AtSQE5 (Laranjeira et al. 2015) was significantly downregulated in the Atdoa10a/b RNAi 4-2 transcriptome, potentially suggesting negative feedback.Furthermore, direct interaction and ubiquitination of AtSQE1 by AtDOA10s need to be confirmed biochemically.A more detailed analysis of the connection between AtDOA10s, AtSQEs, and other components of the sterol synthesis pathway in Arabidopsis will provide further insight into the complexities of sterol homeostasis in plants.
To conclude, our work suggests that DOA10-like E3 ligases, in contrast to their putative yeast and mammalian homologs, do not play a major role in the degradation of Nt-acetylated proteins in Arabidopsis, suggesting that the plant Ac/ N-degron pathway is less highly conserved across kingdoms than other N-degron pathways and that its E3 ligase component(s) await discovery.Furthermore, we uncover conservation of a DOA10-SQE regulatory module across 1.5 billion years of evolution, which suggests that homeostatic mechanisms controlling sterol biosynthesis have ancient origins.
Arabidopsis plants were grown on soil (Levington M3 compost, vermiculite, and perlite; 4:2:1 ratio), in long-day (16 h light at 22°C) or short-day (8 h light at 22°C) conditions.For sterile growth, seeds were surface sterilized (10% v/v bleach), plated onto half-strength Murashige & Skoog (½ MS) medium (1% w/v agar, pH 5.7), stratified at 4°C for 48 h, and grown in long days.ABA (Sigma-Aldrich) was added directly into the ½ MS growth medium to the appropriate concentration.LDAO (Cayman Chemical Company) treatments were carried out on 7-day-old seedlings in water supplemented with 100 µg ml −1 LDAO.Terbinafine hydrochloride (20 µM) (Sigma-Aldrich) was applied to seedlings in the same manner as cycloheximide (see below).All experiments involving chemical treatment of Arabidopsis were conducted at least three times.

Yeast assays
Yeast (Saccharomyces cerevisiae) cells used were homozygous diploid BY4743 cells derived from the S288C strain (Dharmacon yeast KO collection, Horizon Discovery).Yeast were transformed with AtDOA10A, AtDOA10B, and AtNAA20 (At1g03150) in the pAG416GPD-ccdB-EGFP vector (Addgene plasmid #14196, Susan Lindquist) and AtSQE1 in the pAG413GPD-ccdB-HA vector (Addgene plasmid #14238, Susan Lindquist) or the corresponding empty vectors.Transformation was performed by the lithium acetate method: A 1 µL loop of cells was added to 2 µg of expression clone and 100 μL of transformation buffer (33% v/v polyethylene glycol 3350, 0.33 M lithium acetate, 0.66% v/v β-mercaptoethanol).Cells were briefly vortexed, incubated at 37°C for 45 min (200 rpm), then spread on synthetic dropout (DO) media (Formedium), and incubated at 30°C.Non-transformed cells were grown on yeast-extract peptone dextrose (YPD).G418 (Sigma-Aldrich) was added to media used for the growth of mutant strains.For hygromycin treatments, cells were grown overnight in liquid DO media and diluted to an OD 600 of 1.0.A serial dilution of each culture was then spotted onto plates containing DO media with added hygromycin B (TOKU-E) (50-75 µg mL −1 ).

Reverse transcription PCR (RT-PCR) and reverse transcription quantitative PCR (RT-qPCR)
RNA was extracted from snap-frozen samples using an RNeasy minikit (Qiagen) and analyzed by NanoDrop 1,000 spectrophotometer (Thermo Fisher Scientific).Approximately 1.5 µg of RNA was treated with RQ1 DNase (Promega) and cDNA was synthesized using oligo(dT) primers and SuperScript II Reverse Transcriptase (Thermo Fisher Scientific).For semi-quantitative RT-PCR, synthesized cDNA was used in PCR reactions specific to the gene of interest and ACTIN7.RT-qPCR was performed on 45 ng template cDNA using Brilliant III Ultra-Fast SYBR Green QPCR Master Mix with Low ROX (Agilent), using an AriaMx Real-Time PCR System (Agilent).Relative expression was calculated using the 2 −ΔΔCT method (Livak and Schmittgen 2001) normalized to ACTIN7.All primers used are listed in Supplemental Table S1.

Phylogenetic tree construction
Putative DOA10-like protein sequences were identified by BLASTP at NCBI using ScDOA10 as the template sequence.Sequences were aligned in SEAVIEW5 using the Clustal O method, and the phylogenetic tree was constructed using SEAVIEW5 and BIONJ (BIO Neighbor-Joining) method (Gouy et al. 2021), where Poisson zorrection and bootstrap testing is performed with 1,000 iterations.

Histochemical staining
Arabidopsis seedlings were incubated in GUS stain solution (100 mM phosphate buffer (pH 7.0), 2 mM X-gluc (X-GLUC Direct) 0.1% v/v Triton-X-100, 1 mM potassium ferricyanide, 1 mM potassium ferrocyanide) at 37°C for 24 h.Seedlings were cleared and fixed in 3:1 ethanol/acetic acid, mounted onto microscope slides in 50% v/v glycerol, and imaged on a dissecting microscope.At least three biological repeats were conducted for each transgenic line.
Hyperfilm film (Amersham).Band densities were quantified relative to the most intense band following normalization to actin/tubulin using Fiji image analysis software.All stability assays were conducted at least three times.

Accession numbers
RNA-seq data are available at the NCBI GEO database with accession codes GSE236282 (for doa10a/b RNAi 4-2 and its related Col-0 WT, amiNAA10, and Atnaa20 data sets) and GSE161571 (for Col-0 WT data sets related to amiNAA10 and Atnaa20 analyses).Nt-acetylome data are available at the Proteomics Identification Database (PRIDE) with accession code PXD043217.

Figure 1 .
Figure 1.Structure, functional conservation, and phylogeny of Arabidopsis DOA10-like proteins.A) Schematic diagram of Arabidopsis (At) DOA10 proteins compared to the single yeast (Sc) homolog.Key domains and features are shown.B) Percentage identity between Arabidopsis and yeast DOA10 proteins.Values are shown for total protein length and the RING-CH and TD regions.See also Supplemental Fig. S1.C) Transgene-specific RT-PCR (using 5′ and 3′ primer pairs) confirming AtDOA10A/B-GFP expression in Scdoa10Δ mutant yeast cells.D) The Scdoa10Δ yeast mutant is insensitive to hygromycin relative to WT cells.AtDOA10A-GFP can complement this phenotype, whereas AtDOA10B-GFP cannot.Spots represent 10-fold dilutions from left to right.E) Inferred phylogenetic tree of full-length DOA10-like protein homologs identified in various diploid angiosperm species, yeast, and human (HsMARCH6).Two main groups are identified: (i) a DOA10A-like clade, which is split between monocot and dicot lineages, and (ii) a DOA10B-like clade that is comprised of Brassicaceae-derived sequences only.Bootstrap values are shown, and the separate Brassicaceae groupings are highlighted with a blue box.At, Arabidopsis thaliana; Al, Arabidopsis lyrata; Br, Brassica rapa; Cr, Capsella rubella; Os, Oryza sativa; Bd, Brachypodium distachyon; Ac, Ananas comosus; Sl, Solanum lycopersicum; Ca, Capsicum annuum; Gm, Glycine max; Mt, Medicago truncatula; Pt, Populus trichocarpa; Rc, Ricinus communis.Tree not drawn to scale.

Figure 2 .
Figure 2. AtDOA10A and AtDOA10B are broadly expressed in seedlings and localize to the endoplasmic reticulum.A) Histochemical staining of 7and 14-day-old Arabidopsis seedlings expressing pDOA10A/B::AtDOA10A/B-GUS. Scale bar for all images: 1 mm.B) RT-qPCR of endogenous AtDOA10A and B mRNA in different seedlings and adult tissues (note the y axis Log 10 scale).Relative expression levels were calculated through normalization to AtACT7 and are the average of three biological repeats.Horizontal line shows mean.C) anti-GUS immunoblot of microsomal and soluble protein extracts from seedlings expressing pDOA10A/B::AtDOA10A/B-GUS (expected sizes: AtDOA10A-GUS, ∼195 kDa; AtDOA10B-GUS, ∼170 kDa-although both are detected at around 250 kDa).Anti-CNX1/2 (microsomal) and anti-UGPase (soluble) control blots confirming efficacy of the fractionation are shown.D) Confocal images of N. benthamiana leaf pavement cells transiently co-expressing eYFP-AtDOA10A and the ER marker protein AtVMA12-RFP showing co-localization of YFP and RFP signals.Nuclei are stained with DAPI, and chloroplast auto-fluorescence is also shown.Scale bar: 10 µm.

Figure 3 .Figure 4 .
Figure 3. Generation and phenotypic assessment of AtDOA10A and B mutants.A) Schematic of AtDOA10A and B genes, showing 5′/3′ UTR (black) and exons (grey), T-DNA IDs and insertion sites, position of RNAi construct sequences, and RT-qPCR primers used in (B).B) RT-qPCR of endogenous AtDOA10B in the Atdoa10A control line and homozygous RNAi lines 3-7 and 4-2, using primer pairs (P1 and P2) upstream and downstream of the RNAi target sequence.Expression levels were normalized to AtACT7, and expression in the RNAi lines is shown relative to the endogenous levels of DOA10B in the untransformed Atdoa10A mutant.Data are the average of three biological repeats.Horizontal line shows the mean.C) Rosettes of 6-week-old WT, Atdoa10a, Atdoa10b, and Atdoa10a/b RNAi 4-2 lines grown under short days.Images were digitally extracted for comparison.Scale bar: 2 cm.D) 7-and 14-day-old seedlings grown on control or 0.5 µM ABA-supplemented ½ MS plates.Scale bar: 5 mm.

Figure 5 .
Figure 5. Proteolytic turnover of AtSQE by AtDOA10A and AtNAA20 in heterologous yeast degradation assays.A) Anti-HA immunoblot showing steady-state levels of AtSQE1-HA in WT vs Scdoa10Δ yeast cells.Anti-ACTIN bands are shown on the same blot.CBB, Coomassie Brilliant Blue loading control.B) Steady-state protein (immunoblot) and mRNA (RT-PCR) levels of AtSQE1-HA expressed in WT vs Scdoa10Δ ± co-expression with AtDOA10A or AtDOA10B.C) Cycloheximide (CHX) chase of AtSQE1-HA in WT, Scdoa10Δ, and Scnaa20Δ yeast cells (immunoblot and quantified relative density).D) Cycloheximide chase showing that co-expression of AtDOA10A destabilizes AtSQE1-HA in Scdoa10Δ yeast cells (immunoblot and quantified relative density).E) Cycloheximide chase showing that co-expression of AtNAA20 destabilizes AtSQE1-HA in Scnaa20Δ yeast cells (immunoblot and quantified relative density).

Figure 6 .
Figure 6.Impact of N-terminal mutagenesis on AtSQE1 stability suggests indirect effects of Nt-acetyltransferases on protein turnover in yeast.A) Summary of N-terminal (Nt) mutants and predicted respective NAT activities.B) Steady-state protein levels of AtSQE1-HA Nt-variants in WT and Scdoa10Δ yeast cells.C) Cycloheximide (CHX) chase of WT ME-and mutant MP-AtSQE1-HA in WT yeast cells (immunoblot and quantified relative density).D) Cycloheximide (CHX) chase of WT ME-AtSQE1-HA in WT and Scnaa10Δ yeast cells (immunoblot and quantified relative density).E) Cycloheximide (CHX) chase of mutant MA-AtSQE1-HA in WT and Scnaa20Δ yeast cells (immunoblot and quantified relative density).F) Cycloheximide (CHX) chase of mutant MA-AtSQE1-HA in WT and Scnaa10Δ yeast cells (immunoblot and quantified relative density).

Figure 8 .
Figure 8. Summary of the mevalonate (MVA) and sterol synthesis pathways in Arabidopsis: (1) DOA10 negatively regulates SQE1 stability (this study) and positively regulates HMGR activity (Doblas et al. 2013) in plants, yeast, and humans.Yeast and animal SQEs were previously shown to be targets of DOA10 (Foresti et al. 2013; Zelcer et al. 2014), indicating conservation of this regulatory module across three eukaryotic kingdoms.(2) NATA and B were shown to indirectly contribute to AtSQE1 turnover in yeast (this study), but not in Arabidopsis.(3) LDAO, an inhibitor of several downstream enzymatic steps, also negatively regulates AtSQE1 levels via DOA10-independent mechanism(s).(4) Terbinafine, a chemical inhibitor of SQE enzymatic activity (Nowosielski et al. 2011), indirectly promotes accumulation of AtSQE1 (this study), likely through positive feedback.Dashed lines denote indirect effects.