Arabidopsis thaliana ecotype Columbia-0 (Col-0) wild type was used in this study. Arabidopsis T-DNA insertion lines atarre-1 (SALK_094303), atarre-2 (SALK_034426C; Alonso et al., 2003), and the cer1-4 and cer4-4 mutants were obtained from the Arabidopsis Biological Resource Center (ABRC).¹ GABI-KAT T-DNA line atarre-3 (GABI_383G01) was obtained from gabi-kat.de (Kleinboelting et al., 2012). cer3-6 was a gift from Dr. Takuji Wada (RIKEN, Japan). AtARRE overexpression lines in Col-0 background were identified from the snc1-influencing plant E3 ligase reverse (SNIPER) genetic screen (Tong et al., 2017).

Arabidopsis seeds were germinated on Arabidopsis thaliana (AT) medium (Haughn and Somerville, 1986) supplemented with 1% (w/v) agar and appropriate antibiotics for transgene selection. Seven-day-old seedlings were transplanted to soil (Sunshine Mix 4 or 5, SunGro, Canada) supplemented with liquid AT medium and grown in a plant growth chamber at 20°C under continuous light [100μmolm⁻² s⁻¹ of photosynthetically active radiation (PAR)]. Arabidopsis seeds grown for Agrobacterium-mediated transformation were directly spread on the soil supplemented with liquid AT medium at a density of 100 seeds/6″ pot and grown as described above.

Nicotiana benthamiana seeds were sown directly on soil (Sunshine Mix 4 or 5, SunGro, Canada) supplemented with liquid AT medium at a density of 1 seed/3.5″ square pot. Plants were grown under a 14-h light (25°C with 100μmolm⁻² s⁻¹ PAR) and 10-h dark (20°C) cycle. For the transient expression assay, 4- to 5-week-old plants were taken out of the growth chamber and left at room temperature for 3 to 4h before infiltration.

Plant tissues were collected and immediately frozen in liquid nitrogen. RNA was extracted from Arabidopsis leaves, stems, flowers, seedlings, and roots using TRIzol (Thermo Fisher Scientific) according to the manufacturer’s protocol. RNA was isolated from Arabidopsis siliques using a phenol:chloroform:isoamyl extraction and precipitated by lithium chloride and sodium acetate (Wilkins and Smart, 1996). RNA integrity was examined on a 1% standard agarose gel, and RNA was quantified using a NanoDrop 8000 (Thermo Scientific). Genomic DNA was removed by DNase I treatment (New England Biolabs) following the manufacturer’s protocol, and single-stranded cDNA was synthesized from equal amounts of purified RNA using iScript RT Supermix (Bio-Rad). ACTIN1 was used as an internal control. The iQ SYBR Green Supermix (Bio-Rad) was used in 20μl reactions to perform qPCR in an iQ5 Multicolor Real-Time PCR Detection System (Bio-Rad) as specified by the manufacturer. Four technical replicates were performed for each sample, and gene expression levels were analyzed using the Pfaffl method (Pfaffl, 2001).

Standard methods were used for cloning, and all primer sequences are given in Supplementary Table 1. All constructs were confirmed by sequencing.

The pGreenST/35S:HA-AtARRE construct, which was prepared for the SNIPER screen (Tong et al., 2017), was used as the site-directed mutagenesis template. Of the five splice variants known for the AtARRE gene (AT5G66070), AT5G66070.1 was used for the work described here. The 35S:HA-AtARRE^{(H197Y,H200Y)} site-directed mutagenesis construct was generated using primers H197200Y_F and H197200Y_R designed using the one-step site-directed mutagenesis method (Zheng et al., 2004). The PCR amplification was carried out using Phusion High Fidelity Polymerase (Thermo Fisher Scientific). The PCR products were separated by gel electrophoresis, purified using a PCR Purification Kit (BioBasic), and were further treated with restriction enzyme DpnI (New England Biolabs). The mutations in the 35S:HA-AtARRE^{(H197Y,H200Y)} construct were confirmed by sequencing.

The 35S:GFP-CER3 construct was prepared using Gateway cloning (Thermo Fisher) and destination vectors from Nakagawa et al. (2007). The coding sequence of the CER3 gene was amplified from WT cDNA using CER3cDNA_attbF and CER3cDNA_attbR_WSTOP primers and recombined into the entry vector pDONR221. The insert was then transferred into the destination vector pGWB6 to generate pGWB6/35S:GFP-CER3 and into pGWB15 to generate pGWB15/35S:HA-CER3. The CER3 coding sequence without stop codon was also amplified using CER3cDNA_attbF and CER3cDNA_attbR_NoSTOP primers, recombined to pDONR221, and then transferred to the destination vector pGWB5 to generate pGWB5/35S:CER3-GFP.

The 35S:CER1-GFP construct was made and provided by Dr. Hugo Zheng (McGill University, Canada). The coding region of the CER1 gene was subcloned into the vector pVKH18/35S:GFPC (Dean et al., 2007) to produce the C-terminal CER1-GFP fusion under the control of the enhanced 35S promoter. The 35S promoter fragment was then removed from the vector pVKH18/35S:CER1-GFP and replaced with the CER6 promoter using HindIII and XbaI to generate pVKH18/CER6p:CER1-GFP. The construct pBIN/35S:HDEL-mCherry was provided by Dr. Mathias Schuetz (Nelson et al., 2007), and the construct pGreenST/35S:HA-SNIPER2 was described previously (Wu et al., 2020).

To generate the construct AtARREp:GUS, a fragment of 808bp immediately upstream of the putative AtARRE start codon, which includes the 5′ UTR of AtARRE, as well as 3′ UTR and the last intron of the previous gene, was amplified from WT genomic DNA using LP_attb1_AtARRE and RP1_attb2_AtARRE and recombined into pDONR221 before being introduced into pGWB3 (Nakagawa et al., 2007; Vincent et al., 2018) using GATEWAY cloning (Thermo Fisher).

Standard methods were used for cloning, and all primer sequences are given in Supplementary Table 1. All constructs were confirmed by sequencing.

To generate the construct pET28b/AtARRE-HIS for the in vitro ubiquitination assay, a 326bp fragment of coding sequence downstream of the transmembrane domains and upstream of the stop codon of AtARRE was amplified from WT cDNA using AtARRETMdel_F_EcoRI_28b and AtARRETMdel_R_SalI. This PCR product was ligated into the pET28b vector using EcoRI and SalI restriction sites to generate pET28b/AtARRE-HIS.

To reconstitute the plant ubiquitination cascade in Escherichia coli, Duet expression vectors (kindly provided by Dr. Dongping Lu, Chinese Academy of Science, China) pCDFDuet/MBP-ABI3-HA-AtUBA1-S, pCDFDuet/AtUBA1-S, pACYCDue/AIP2-Myc-UBC8-S, and pET28a/FLAG-UBQ were used to generate target co-expression constructs (Han et al., 2017). A 326bp fragment of coding sequence downstream of the transmembrane domains and upstream of the stop codon of AtARRE was amplified from WT cDNA and ligated into the BamHI and StuI-digested pACYCDue/AIP2-Myc-UBC8-S vector to generate pACYCDue/AtARRE-Myc-UBC8-S. An 837bp fragment of coding sequence downstream of the transmembrane domains and upstream of the stop codon of CER1 was amplified from WT cDNA was ligated into the EcoRI and StuI-digested pCDFDuet/MBP-ABI3-HA-AtUBA1-S vector to generate pCDFDuet/MBP-CER1-HA-AtUBA1-S.

To produce transgenic lines for E3 ubiquitin ligase activity test, degradation assay, and GUS assay, 35S:HA-AtARRE^{(H197Y,H200Y)}, CER6p:CER1-GFP, AtARREp:GUS, and AtARREp:AtARRE-GUS were introduced into Agrobacterium tumefaciens GV3101 cells carrying the pMP90 Ti plasmid. The pGreenST plasmid 35S:HA-AtARRE^{(H197Y,H200Y)} was co-transformed with the helper plasmid pSOUP (Hellens et al., 2000). Transformation of WT or cer1-4 plants was carried out using the floral spray method (Chung et al., 2000). T1 transgenic seeds were harvested and screened on AT medium supplemented with 1% (w/v) agar and appropriate antibiotics.

Transient expression in N. benthamiana was carried out using 4- to 5-week-old plants. Agrobacterium cultures were grown overnight in 3ml of LB medium under antibiotic selection and diluted 1/20 in LB medium with antibiotics and 50μm acetosyringone and incubated for a further 3–5h. During this time, plants were taken out of the growth chamber and left at room temperature before infiltration. Cultures were centrifuged and resuspended in resuspension medium (4.43g/L MS, 10mm MES, and 150μm acetosyringone) at an optical density of 0.6 at A₆₀₀. For co-expression of multiple constructs, suspensions were mixed in equal ratios. Agrobacterium suspension mixtures were infiltrated using a 1-ml syringe into the abaxial side of the N. benthamiana leaves. A permanent marker was used to mark the infiltrated area on the leaf. Infiltrated plants were incubated at room temperature for 48h, and then, leaf samples were collected for microscopic imaging and/or protein extraction.

Cuticular wax extraction was performed using the method described by Haslam and Kunst (2013b). Briefly, the top 10cm of 4- to 6-week-old inflorescence stems were cut and photographed to allow stem surface area to be calculated by measuring the number of pixels of the two-dimensional area in Photoshop (Adobe), converting the values to cm², and multiplying by π. After imaging, stems were submerged for 30s in chloroform containing 10μg tetracosane as an internal standard. After wax extraction, chloroform was evaporated under a stream of nitrogen gas and wax components were silylated in 10μlN, O-Bis (trimethylsilyl) trifluoroacetamide (BSTFA; Sigma), and 10μl pyridine for 1h at 80°C. After derivatization, the solvent was evaporated under nitrogen gas and waxes were re-dissolved in 30μl of chloroform for GC analysis. Samples were analyzed on an Agilent 7890A gas chromatograph equipped with a flame ionization detector (GC-FID) using an HP1 column (Agilent) in a 2.7:1 split mode with H₂ as the carrier gas at a flow rate of 30ml/min. The gas chromatography program used was as follows: oven temperature was set at 50°C for 2min, raised by 40°C/min to 200°C and held for 1min, and then raised by 3°C/min to 320°C and held for 15min. Wax components were identified by comparing their retention times with those of the internal standards. Four biological replicates were processed for each line.

Fluorescence signals of transiently expressed constructs in N. benthamiana were detected using a Perkin Elmer Ultraview VoX Spinning Disk Confocal Microscope. N. benthamiana leaf discs were mounted in distilled water and immediately imaged using a glycerol immersion lens. GFP was excited using a 488nm laser with an 515/30nm emission filter, and mCherry was excited using a 561nm laser with an 595/50nm emission filter. Confocal images were processed using the Volocity software (Perkin Elmer). GFP signal in infiltrated N. benthamiana leaf was also observed using a Nikon Eclipse 80i Scanning Laser Confocal Microscope excited with a 488nm laser with an 515/30nm emission filter.

For scanning electron microscopy (SEM), segments from the apical 1cm of dry stems were mounted onto stubs and sputter-coated with gold particles for 10min at 40mA in an SEM Prep 2 sputter coater (Nanotech). The coated samples were viewed using an S4700 field emission SEM (Hitachi) with an accelerating voltage of 5kV and a working distance of 12mm.

Tissues at different developmental stages from transgenic lines expressing AtARREp:GUS constructs were immersed in GUS staining buffer {100mM Na-phosphate, 10mM EDTA, pH 7.0, 0.5mM K₃[Fe(CN)₆], 0.5mM K₄[Fe(CN)₆], 0.1% (v/v) Triton X-100, and 1mM 5-bromo-4-chloro-3-indolyl-β-D-glucuronide (X-gluc)} and incubated for 1 to 3h or overnight. The reaction was stopped by removing the GUS buffer and adding the 70% (v/v) ethanol. Chlorophyll was removed by incubating samples in 70–90% (v/v) ethanol before samples were examined under a Nikon SMZ18 Digital Microscope (Nikon, Japan).

Plant tissues were ground in liquid nitrogen, and total proteins were extracted in buffer containing 50mM Tris-HCl, pH 7.5, 150mM NaCl, 1mM EDTA, 10% (v/v) glycerol, 1% Triton X-100, 1mM PMSF, and 1X Halt^™ protease inhibitor cocktail (Thermo Fisher Scientific). After centrifugation at 18,000g for 20min at 4°C, the supernatant was transferred to a new tube and the concentration of protein extract was determined using the Bradford reagent (Bio-Rad).

For SDS-polyacrylamide gel electrophoresis (SDS-PAGE), 4X SDS loading buffer (200mM Tris-HCl, pH 6.8, 8% (w/v) SDS, 0.4% (w/v) bromophenol blue, 40% glycerol, and 400mM DTT) was added to solubilized protein samples, and 10–35μl of each protein sample was separated on a 10% acrylamide gel with 1% SDS at 200V constant voltage for 50–60min before being transferred to nitrocellulose membrane using a semi-dry blotting system (Bio-Rad) with Bjerrum Schafer-Nielsen buffer (Bio-Rad). Transfer was carried out at a constant voltage of 15V for 50min before the membrane was stained with Ponceau S, imaged, washed, and then blocked with 5% skim milk powder in Tris-buffered saline with 0.1% tween 20 (TBS-T). For immunoblotting, membranes were incubated with primary antibody for 1h at room temperature. Primary antibodies used were anti-GFP (dilution 1:5,000; mouse IgG; Roche), anti-HA (dilution 1:2,500; rat IgG; Roche), anti-HIS (dilution 1:1,000; mouse IgG; Santa Cruz Biotechnology), anti-FLAG (dilution 1:5,000; mouse IgG; Sigma), anti-GST (dilution 1:1,000; rabbit IgG; Sigma), anti-Myc (dilution 1:1,000; Invitrogen), and anti-Ub (dilution 1:1,000; mouse IgG; Sigma). Membranes were then washed three times for 10min each wash with TBS-T and then incubated with appropriate secondary antibodies, including anti-rabbit (dilution 1:10,000; Santa Cruz Biotechnology), anti-mouse (dilution 1:25,000; Santa Cruz Biotechnology), and anti-rat (dilution 1:10,000; Santa Cruz Biotechnology), for 1h at room temperature. The membrane was washed three times as above with TBS-T before horseradish peroxidase was detected with the ECL Prime western blotting detection kit (GE).

Plant-derived protein degradation assays were performed as described in Wang et al. (2009), with modifications as follows. Total proteins were extracted from 8-day-old CER6pro:CER1-GFP/cer1-4 transgenic seedlings and quantified. 80μl protein extracts were then incubated with or without 40μM MG132 (Sigma) at 30°C. Samples were taken at select time points, and the reaction was stopped by adding 5μl 4X SDS loading buffer. CER1-GFP protein abundance in each sample was determined by immunoblotting using anti-GFP antibody.

In vitro ubiquitination assays were performed as described in Zhao et al. (2013), with modifications as follows. The plasmid pET28b/AtARRE-HIS was transformed into E. coli strain Rosetta^™2 (DE3) for protein production. A 500ml culture was grown in Terrific Broth (TB) medium [1.2% (w/v) tryptone, 2.4% (w/v) yeast extract, 0.4% (v/v) glycerol, 100mM K-PO₄] until the exponential phase (OD₆₀₀ =0.6–1.0) before protein production was induced by adding 0.5mM Isopropyl β-D-1-thiogalactopyranoside (IPTG). After growth overnight at 16–18°C, cells were collected by centrifugation at 6,000g for 5min and frozen in liquid nitrogen and stored at −80°C. Lysis buffer [50mM NaPO₄, pH 7.5, 200mM NaCl, 0.1% (v/v) Triton X, 5% glycerol, 1mM PMSF, 1X Halt^™ protease inhibitor cocktail, and 1mg/ml lysozyme] was added to the frozen sample pellets, thawed at 37°C for 1min, and resuspended. Lysate was cleared by centrifugation at 14,000g for 15min at 4°C followed by filtration through a 0.45 μm filter. The AtARRE-HIS recombinant proteins were purified using HisPur Ni-NTA Resin (Thermo Fisher) according to the manufacturer’s protocol. Purified recombinant proteins AtUBA2-His and GST-AtUBC8 were kindly provided by Dr. Oliver Xiao'ou Dong (Dong et al. 2018).

In vivo ubiquitination assays in bacteria were carried out using the system described by Han et al. (2017). E. coli strain BL21 (DE3) containing different combinations of the expression vectors were grown in 2ml of TB liquid medium with appropriate antibiotics at 37°C. When the culture A₆₀₀ nm reached 0.4–0.6, 0.5mM IPTG was added to induce the recombinant protein expression. After induction, bacteria were further grown at 28°C for 10–12h, stored at 4°C overnight, and then harvested from 300μl of culture by centrifugation at 12,000g for 5min. The pellets were resuspended in 100μl 1x SDS loading buffer and boiled at 95°C for 5min followed by immunoblotting.

To identify novel E3 ligases involved in plant immunity, a SNIPER genetic screen has been carried out (Tong et al., 2017). In this screen, E3-ligase encoding genes induced during plant defense were overexpressed in the wild-type background. Unexpectedly, a number of independent transgenic plants with glossy bright green stems were uncovered among the T1 progeny, suggestive of altered cuticular wax accumulation (Figure 1A). In these plants, the AtARRE/At5g66070 gene encoding a RING-type E3 ubiquitin ligase was expressed under the control of the cauliflower mosaic virus (CaMV) 35S promoter (Supplementary Figures S1A,B). Wax analysis of three representative AtARRE overexpression (AtARREOX) lines by gas chromatography demonstrated that they accumulated only 10–50% of the WT inflorescence stem wax and only ~65% of the WT leaf wax (Figures 1B,C). The stem wax phenotype was further evaluated by scanning electron microscopy (SEM); wild-type stem surface was densely and uniformly covered with column-, vertical plate-, and rod-shaped wax crystals, whereas AtARREOX lines displayed considerably lower density of all types of wax structures (Figure 1D).

The wax deficiency uncovered in the AtARREOX lines prompted us to examine the wax phenotypes of atarre mutants. We obtained three T-DNA insertion lines of AtARRE in the Col-0 ecotype and determined AtARRE gene expression in mutant alleles by qPCR. Even though we detected reduced levels of AtARRE transcript in all three T-DNA lines (Supplementary Figures S1A,B), we found no major differences in the total stem wax load or composition with respect to the wild type (Supplementary Figures S1C,D).

To further investigate the role of AtARRE in cuticular wax biosynthesis, we carried out a detailed analysis of wax composition of AtARREOX lines. We found that amounts of all stem wax components were altered in AtARREOX plants in comparison with the wild type, and detected considerable changes in their relative proportions. In particular, there was a prominent decrease in absolute amounts of alkane pathway-derived compounds that could be attributed primarily to C29 alkanes (68–98.7% decrease), C29 ketones (50–97.6% decrease), and C29 secondary alcohols (56–97% decrease). Conversely, the amounts of fatty acids on AtARREOX stems were higher than in the wild type (115–167% increase) and so was the relative proportion of fatty acids, aldehydes, and primary alcohols (Figure 1E, Supplementary Figure S1E). We also examined the wax composition of AtARREOX rosette leaves. As observed with stem wax, leaf wax also contained significantly reduced amounts of C29 and C31 alkanes relative to the wild type, but also increased amounts of C33 alkanes (Figure 1F). In addition, we detected a major increase in C24 and C26 fatty acids, as well as C26 primary alcohols (Figure 1F). Collectively, these data suggest that wax production by the alkane-forming pathway is impaired in AtARRE overexpressors.

AtARREOX lines with the most severe wax deficiency displayed additional phenotypes including abnormal organ morphogenesis, dwarfism, organ fusions, and reduced fertility over multiple generations (Figure 2). For example, the inflorescence stems of AtARRE overexpressors were considerably shorter than the wild type. Additionally, these plants also exhibited organ fusions between flower buds, flowers, and siliques, as well as between flowers and leaves (Figures 2C,D). Reduced fertility was also often detected. In most cases, fertility could be restored by growing plants under high humidity, except in individuals with severe floral organ fusions (Supplementary Figure S2). Similar phenotypes have previously been reported for several Arabidopsis wax-deficient eceriferum mutants (Koornneef et al., 1989).

AtARRE (AT5G66070) encodes a predicted polypeptide of 221 amino acids with a molecular mass of 27kDa containing three transmembrane domains located at the N-terminus and a RING domain located at the C-terminus (Figure 3A). RING domain proteins act as E3 ligases by binding to an E2-ubiquitin thioester and catalyzing ubiquitin transfer (Deshaies and Joazeiro, 2009). Whether the RING domain-related E3 ubiquitin ligase activity of the AtARRE protein is required for its function in cuticular wax metabolism is not known. RING domain E3 ligases are known to undergo self-ubiquitination in the absence of their native substrate (Lorick et al., 1999). We used this feature of RING E3 ligases to determine whether AtARRE has E3 ligase activity. Unfortunately, the insolubility of the full-length AtARRE protein upon expression in E. coli prohibited purification of enough protein for the self-ubiquitination assay. Therefore, an AtARRE protein fragment without the N-terminal transmembrane domains was used to produce the recombinant AtARRE-HIS protein. Incubation of the purified AtARRE-HIS recombinant protein with E1, E2, ubiquitin, and ATP resulted in a laddering pattern characteristic of ubiquitination on a protein blot when anti-HIS antibodies were used for AtARRE-HIS detection. This laddering is indicative of a range of molecular weights for AtARRE-HIS as it carries ubiquitin chains of different lengths (Figure 3B). Such a laddering pattern was also detected when anti-Flag antibodies that were used for FLAG-Ub detection of Ub chains bound to AtARRE-HIS. Thus, AtARRE exhibits E3 ligase activity and undergoes self-ubiquitination in vitro. The self-ubiquitination of AtARRE was not observed when E1, E2, or ubiquitin were omitted from the assays.

Conserved Cys and His residues in the RING domain are critical for the E3 ligase activity (Deshaies and Joazeiro, 2009). Their substitution disrupts the RING domain and results in a dominant-negative form of E3 ligase predicted to confer the same phenotype as the loss of E3 ligase function. To further verify whether AtARRE E3 ligase activity is required for its function, we replaced the conserved AtARRE RING domain His-197 and His-200 with Tyr residues by site-directed mutagenesis and expressed the modified protein in wild-type plants. In contrast to AtARRE overexpression which caused wax deficiency, overexpression of the double mutant AtARRE^(H197YH200Y) protein had no effect on the stem wax load (Figure 3C). Thus, the E3 ligase activity of AtARRE is required for its function in stem wax deposition.

E3 ligase-mediated ubiquitination of proteins in most cases results in their degradation by the 26S proteasome. Because the most conspicuous result of AtARRE overexpression was reduced cuticular wax accumulation on Arabidopsis inflorescence stems, we hypothesized that AtARRE may act as a negative regulator of wax deposition by ubiquitinating, and thus targeting for degradation, a key player involved in wax biosynthesis. If this is the case, identifying the ubiquitination substrate of AtARRE is critical for determining its biological function. As a first step in uncovering potential candidate ubiquitination substrates, we compared stem wax compositional changes of AtARRE overexpressors with those of ecerifierum (cer) Arabidopsis mutants caused by loss-of-function mutations in wax biosynthetic genes (Figure 4, Supplementary Figure S3). AtARRE overexpressors displayed dramatically reduced alkane, secondary alcohols, and ketone levels on their stem surfaces, similar to null mutants disrupted in CER1 and CER3 genes required for the production of waxes by the alkane-forming branch of wax biosynthesis, but not the mutants with lesions in the CER4 gene required for the production of waxes by the acyl reduction pathway. The distinguishing feature between cer1 and cer3 is that cer1 has slightly increased amounts of aldehydes and reduced primary alcohol levels, whereas cer3 exhibits a major reduction in aldehydes and similar amounts of primary alcohols to the wild type. Thus, the wax profile of AtARREOX lines is most similar to that of the cer1 mutant. Besides cuticular wax changes, some AtARREOX lines additionally display reduced plant height and reduced fertility previously described for the cer1-1 and cer3-1 mutant alleles (Koornneef et al., 1989), and organ fusion phenotypes characteristic of cer3, but not cer1 mutants (Aarts et al., 1995; Chen et al., 2003; Bourdenx et al., 2011).

To determine whether CER1 is subjected to 26S proteasome-dependent degradation, we performed a modified cell-free degradation assay. For this purpose, we made CER6pro:CER1-GFP/cer1-4 transgenic lines in which the CER6pro:CER1-GFP transgene complemented the cer1-4 wax deficiency (Supplementary Figure S4A). Total proteins extracted from 8-day-old CER6pro:CER1-GFP/cer1-4 seedlings were incubated with or without the 26S proteasome inhibitor MG132 for 90min and protein levels determined by immunoblotting. The CER1-GFP amounts decreased rapidly in the absence of MG132, but in the presence of MG132, the levels of CER1-GFP remained notably higher over time, suggesting that the 26S proteasome is involved in CER1 proteolysis (Figure 5A).

We next carried out an Agrobacterium-mediated transient expression assay in N. benthamiana leaves. Agrobacterium cell cultures expressing 35Spro:CER1-GFP and 35Spro:HA-AtARRE were co-infiltrated on one half of a N. benthamiana leaf. Cell cultures expressing 35Spro:CER1-GFP and 35Spro:HA empty vector or 35Spro:CER1-GFP and the 35pro:HA-SNIPER2 vector containing the SNIPER2 E3 ligase gene not involved in wax deposition were co-infiltrated symmetrically on the other half of the same leaf as negative controls (Supplementary Figure S4B). 35Spro:HDEL-mCherry was also included in each sample as a marker for ER visualization. In N. benthamiana cells transformed with CER1-GFP and 35Spro:HA empty vector, the GFP signal could be easily detected after 48h, persisted past 72h and was undetectable 96h after infiltration (Supplementary Figure S4C). Bright GFP fluorescence was also detected in cells co-expressing CER1-GFP and the SNIPER2 E3 ligase control 3days post-infiltration, but not in those cells co-expressing CER1-GFP and AtARRE (Figure 5B). In contrast, similar intensity of mCherry fluorescence from the ER-localized HDEL-mCherry marker was detected in all infiltrated regions on both sides of the leaf (Figure 5B). Immunoblot analysis confirmed that CER1-GFP protein level was much lower in the presence of AtARRE than in the negative control sample expressing SNIPER2 E3 ligase (Figure 5D). Unlike the native AtARRE protein, co-expression of CER1-GFP and the AtARRE^(H197YH200Y) protein with mutated RING domain did not affect CER1 protein levels, indicating that catalytic activity of AtARRE is required for CER1 degradation (Figures 5C,D). When the substrate CER1 was replaced with CER2, a component of VLCFAs elongation machinery (Haslam et al., 2012), GFP tagged CER2 fluorescence signal intensity was found to be indistinguishable in the presence and absence of AtARRE. Immunoblot results were consistent with microscopy data (Supplementary Figures S4D,E) and demonstrate that AtARRE specifically targets CER1 for degradation by the 26S proteasome.

The AtARRE-dependent degradation of CER1 was further verified in stable transgenic lines of Arabidopsis. Plants harboring the CER1-GFP transgene were crossed with the wild type, AtARREOX, and AtARREOX^(H197YH200Y) lines, and the abundance of the CER1-GFP in F1 progeny was examined by immunoblotting. Whereas CER1-GFP transcript accumulation was similar in the F1 progeny from all the crosses, the CER1-GFP protein level in the AtARREOX lines was much lower than observed in the wild type and AtARREOX^(H197YH200Y) lines (Figure 5E). Collectively, these results confirm that AtARRE promotes the degradation of CER1, suggesting that CER1 is the ubiquitination substrate of AtARRE.

To directly test whether CER1 is a ubiquitination substrate of the AtARRE E3 ligase, we performed an in vivo ubiquitination assay in a heterologous E. coli system that expresses the Arabidopsis ubiquitination cascade (Han et al., 2017). In this experiment, recombinant ubiquitination components E1 (AtUBA1-S), E2 (AtUBC8-S), E3 (AtARRE-Myc), ubiquitin (His-Flag-AtUBQ10), and the presumed ubiquitination substrate (MBP-CER1-HA) were co-expressed in E. coli (Figure 6A), and bacterial lysates were analyzed by immunoblotting. Our results show that in the presence of all ubiquitination components, a smear indicative of CER1 ubiquitination can be detected by an anti-HA antibody (Figure 6B). Using an anti-Myc antibody, AtARRE-Myc recombinant protein also shows a laddering pattern indicative of AtARRE self-ubiquitination. These data support the in vitro ubiquitination assay results that AtARRE is an active E3 ligase and demonstrate that it can ubiquitinate CER1 in vivo.

Even though stem wax profiles of AtARRE overexpressors are more similar to the cer1 than the cer3 mutant, the overexpression lines also exhibit organ fusions previously detected in cer3, but not in cer1 mutants. Because CER1 and CER3 proteins are highly related and share 35% sequence identity (Bernard et al., 2012), it is tempting to speculate that in addition to CER1, CER3 may also be a target of the AtARRE-mediated ubiquitination and UPS proteolysis resulting in the dual cer1- and cer3-like phenotypic features of AtARRE overexpression lines. To investigate whether this is the case, we tested whether AtARRE affects the protein levels of CER3-GFP in N. benthamiana leaf epidermal cells. Agrobacterium cells expressing 35Spro:CER3-GFP, the 35Spro:HA-AtARRE, and the 35Spro:HDEL-mCherry transgenes were co-infiltrated on one half of a N. benthamiana leaf, while Agrobacterium cells harboring 35Spro:CER3-GFP, 35Spro:HA-SNIPER2, and 35Spro:HDEL-mCherry were co-infiltrated on the other side of the same leaf as a negative control. Similar to CER1-GFP, AtARRE promoted CER3-GFP degradation as indicated by considerably reduced CER3-GFP fluorescence in the presence of transiently co-expressed AtARRE in comparison with the control co-expressing CER3-GFP with HA-SNIPER2 E3 ligase. By contrast, the signal of the internal control HDEL-mCherry was prominent and indistinguishable on both halves of the leaf (Figure 7). These results demonstrate that CER3 is likely also the AtARRE ubiquitination target. We tried to confirm that AtARRE controls the levels of CER3 in transgenic lines of Arabidopsis, but consistently failed to detect CER3-GFP protein on immunoblots.

To obtain clues regarding the functional significance of the AtARRE-mediated protein degradation of wax biosynthetic enzymes, we investigated the transcription profile of AtARRE gene in different plant tissues using qRT-PCR (Figure 8A). The AtARRE gene was expressed in all tissues examined, with higher expression in rosette leaves and lower expression in flowers and siliques. To further determine the AtARRE expression pattern in different cell types and at different Arabidopsis developmental stages, an 808bp fragment of genomic sequence immediately upstream of the AtARRE coding region was fused to the β-glucuronidase (GUS) reporter gene, and this AtARREpro:GUS reporter was transformed into wild-type Arabidopsis. GUS activity was examined in tissue samples from mature plants and in developing seedlings of ten independent transgenic lines.

In mature plants, high GUS activity was detected in the fully expanded rosette leaves, sepals, pollen grains, and cauline leaves (Figures 8B–F). These results are consistent with the published RNA-seq data showing high AtARRE expression in mature leaves and sepals (Klepikova et al., 2016). Strong GUS signal was also detected in, but not specific to, the epidermal cells of rosette leaves, as well as trichomes, which are specialized epidermal cells (Figure 8G). Surprisingly, no expression of AtARRE was detected in the epidermal cell layer of the stem (Figures 8H,I). This may be due to the fact that the 5′ promoter fragment used in the AtARREpro:GUS construct does not contain the regulatory element required for the AtARRE expression in stem epidermal cell layer. In fact, according to published RNA-seq and microarray data, AtARRE shows the highest expression level in the senescent first internode of the stem, and higher expression was detected in the epidermal cell layer at the bottom of the inflorescence stem compared to the top of the stem (Suh et al., 2005; Klepikova et al., 2016). During seedling development, GUS activity was first detected in roots in 3-day-old seedlings and cotyledons at 4days after imbibition (Supplementary Figure S5), but the GUS staining was much more pronounced in both organs in 5-day-old seedlings.

The observed expression profile fits the role of AtARRE as a negative regulator of wax biosynthesis in tissues that exhibit no or low wax production, such as mature cotyledons in older developing seedlings, as well as fully expanded rosette leaves and older internodes at the bottom of the stem in mature plants. Unexpectedly, AtARRE was also found to be expressed in roots, even though CER1 and CER3 genes encoding AtARRE ubiquitination substrates are not expressed in roots or are expressed at a very low level, respectively (Bourdenx et al., 2011).

Because AtARRE was discovered in a reverse genetic screen conducted to identify novel plant immunity-related E3 ligases, we were interested in determining whether AtARRE is induced upon exposure to pathogens. An earlier study has shown that AtARRE is upregulated after treatment with chitin, a potent elicitor of plant defense responses (Libault et al., 2007). Our qRT-PCR analysis revealed that elicitation with flg22, a peptide derived from bacterial flagellin, also results in a major increase in AtARRE expression (Figure 9A). In addition, infiltration of plants with the type III secretion deficient bacterial strain Pseudomonas syringae pv. tomato (P.s.t.) DC3000 hrcC⁻ and the virulent bacterial strain P. syringae pv. maculicola (P.s.m.) ES4326 (Figures 9B,C) strongly upregulated AtARRE expression. Collectively, these data suggest that AtARRE may be involved in regulating cuticular wax biosynthesis in response to pathogen attack.