Arabidopsis RING Peroxins are E3 Ubiquitin Ligases that Interact with Two Homologous Ubiquitin Receptor Proteins


Tel: +1 517 432 4620; Fax: +1 517 353 9168; E-mail:


Peroxisomes are essential eukaryotic organelles that mediate various metabolic processes. Peroxisome import depends on a group of peroxisome biogenesis factors called peroxins, many of which are evolutionarily conserved. PEX2, PEX10, and PEX12 are three RING-finger-domain-containing integral membrane peroxins crucial for protein import. In yeast (Saccharomyces cerevisae), RING peroxins act as E3 ligases, facilitating the recycling of the peroxisome import receptor protein PEX5 through ubiquitination. In plants, RING peroxins are essential to plant vitality. To elucidate the mode of action of the plant RING peroxins, we employed in vitro assays to show that the Arabidopsis RING peroxins also have E3 ligase activities. We also identified a PEX2-interacting protein, DSK2b, which is a member of the ubiquitin receptor family known to function as shuttle factors ferrying polyubiquitinated substrates to the proteasome for degradation. DSK2b and its tandem duplicate DSK2a are localized in the cytosol and the nucleus, and both interact with the RING domain of PEX2 and PEX12. DSK2 artificial microRNA lines did not display obvious defects in plant growth or peroxisomal processes, indicating functional redundancies among Arabidopsis ubiquitin receptor proteins. Our results suggest that Arabidopsis RING peroxins can function as E3 ligases and act together with the ubiquitin receptor protein DSK2 in the peroxisomal membrane-associated protein degradation system.


Peroxisomes are single membrane-bounded organelles that are ubiquitous in eukaryotes. They mediate an array of biochemical reactions and are crucial players in lipid metabolism and detoxification reactions (Islinger et al. 2010). In plants, peroxisome functions have also been implicated in the glyoxylate cycle, photorespiration, hormone metabolism (indole-3-butyric acid, IBA; jasmonic acid, JA), and other metabolic functions (Kaur et al. 2009; Hu et al. 2012). The significance of peroxisomes is underscored by the numerous human diseases and lethal plant phenotypes caused by peroxisomal deficiencies (Steinberg et al. 2006; Hu et al. 2012).

Peroxisomes are metabolically plastic, remodeling their constituent enzyme complement depending on cell type, developmental cues, and prevailing environmental conditions (Hayashi and Nishimura 2006; Islinger et al. 2010). The changes in peroxisome proteome are accomplished by virtue of the dynamic import machinery, which comprises a set of proteins called peroxins (PEX), many of which are conserved from fungi, animals to plants (Distel et al. 1996; Hu et al. 2012). Peroxisome matrix proteins, most of which are distinguished by the presence of a peroxisome targeting signal (PTS), are encoded in the nucleus and imported into the peroxisome post-translationally. There are two kinds of PTS: PTS1 is a C-terminal tripeptide, while PTS2 is an N-terminal cleavable nonapeptide (Platta and Erdmann 2007). PEX5 and PEX7 are cytosolic receptors for PTS1- and PTS2-containing proteins, respectively. In plants, the PTS2-bound PEX7 protein further binds cooperatively with PEX5 to form an import-competent cargo-receptor complex (Hu et al. 2012). The cargo-receptor complex is then ferried to the membrane-docking complex consisting of PEX14 and PEX13. This event is followed by translocation and finally export of the receptor, which is then ready for another round of protein import (Rucktaschel et al. 2011).

Translocation is accomplished by the consecutive activities of the RING finger complex (PEX2, PEX10, PEX12), the ubiquitin-conjugating enzyme (UBC) complex (PEX4, PEX22) and two AAA ATPases (PEX1, PEX6). Although the post-docking events are not very well understood, it is known that the three RING peroxins are vital for peroxisome import to occur (Rucktaschel et al. 2011). These RING peroxins are thought to form an importomer to aid matrix protein import, but their exact mode of action is unknown. In both yeast (Saccharomyces cerevisae) and mammals, the recycling of the PTS1 receptor PEX5 is contingent upon monoubiquitination (Kragt et al. 2005; Carvalho et al. 2007; Platta et al. 2007; Grou et al. 2009; Okumoto et al. 2011). In yeasts, the E2 activity is imparted by PEX4, while in mammals the E2D family of UBCs catalyzes PEX5 monoubiquitination (Platta et al. 2007; Grou et al. 2008). Furthermore, if recycling is compromised, the yeast PEX5 undergoes polyubiquitination and proteasomal-based degradation (Platta et al. 2004; Kiel et al. 2005; Kragt et al. 2005). In this context, the yeast RING peroxins have been demonstrated to have ubiquitin ligase activity, wherein PEX12 functions as a monoubiquitin ligase while PEX2 and PEX10 contain polyubiquitinating activities (Williams et al. 2008; Platta et al. 2009).

So far, there is no direct evidence to support the notion that PEX5 in plants has to be ubiquitinated prior to export. It has been noted that the PEX5 Cys residue that undergoes ubiquitination is conserved in plants, and that the import mutant pex6 has lower steady state levels of PEX5, hinting at similar mechanisms for PEX5 recycling in plants (Zolman and Bartel 2004). Similarities such as the presence of UBCs, RING ligases and AAA ATPases, have also been drawn between the peroxisome import machinery and Endoplasmic Reticulum-Associated protein Degradation (ERAD), a quality control system that removes misfolded proteins from the endoplasmic reticulum (ER) (Schluter et al. 2006). In support of this model, Arabidopsis pex5, pex4 pex22 double and pex6 mutants show enhanced stability of two transiently expressed enzymes of the glyoxylate cycle, isocitrate lyase and malate synthase, during the developmental transition of seedling peroxisomes to leaf peroxisomes (Lingard et al. 2009). This study suggests that these peroxins may have a role in the regulated removal of obsolete or damaged proteins from within the peroxisome.

In yeast, null mutants of the RING peroxins are viable, but the peroxisomes are not import-competent (Platta and Erdmann 2007). In mammals, mutations in any of the RING peroxins result in fatal genetic diseases such as the Zellweger spectrum disorders (Steinberg et al. 2006). In Arabidopsis, null mutants of the RING peroxins have embryo-lethal phenotypes, implying the essential functions of these proteins for survival (Hu et al. 2002; Schumann et al. 2003; Sparkes et al. 2003; Fan et al. 2005). It is also speculated that the plant RING peroxins have novel functions apart from their role in matrix protein import. For example, a gain-of-function allele of PEX2 (ted3) suppresses the phenotypes of the photomorphogenic mutant det1, while RNAi lines of PEX10 have defects in cuticular wax accumulation (Hu et al. 2002; Kamigaki et al. 2009). In addition, plants overexpressing a dysfunctional PEX10 contained deformed peroxisomes with reduced contact with chloroplasts, and exhibited impaired photorespiration. This change in peroxisome morphology was linked to the TLGEEY motif in the N terminus of PEX10, and the RING domain was implicated in mediating inter-organellar contact (Schumann et al. 2007; Prestele et al. 2010).

To elucidate the molecular function of the RING peroxins in plants, we first employed in vitro assays to demonstrate that all three Arabidopsis RING peroxins act as ubiquitin ligases. Using a yeast two-hybrid (Y2H) screen with the RING domain of PEX2 as bait, we identified DSK2b, a ubiquitin receptor protein that links the substrate ubiquitination process and proteasomal degradation events. The interaction was confirmed with in vitro pull-down assays, and structural motifs responsible for the interaction were dissected. Both DSK2b and its closely-related paralog DSK2a could interact with PEX12 in addition to PEX2, and both are localized to the cytosol as well as the nucleus. Reducing the expression of the DSK2 genes by artificial microRNA (amiRNA) did not cause obvious defects in peroxisome function, suggesting functional redundancy between members of the ubiquitin receptor family in Arabidopsis.


Arabidopsis RING peroxins possess E3 ligase activity

The majority of the RING proteins, including the yeast RING peroxins, function as E3 ligases (Deshaies and Joazeiro 2009). To investigate if the Arabidopsis RING peroxins possess E3 ligase activity, we cloned, expressed, and purified maltose binding protein (MBP) fusions of the PEX2/10/12 RING domains, all of which reside at the cytoplasmic C terminus of the proteins (Figure 1A, B). The cellular ubiquitination cascade was reconstituted using an in vitro ubiquitination assay comprised of a wheat E1 ubiquitin activating enzyme, the human E2 ubiquitin conjugating enzyme UBCH5b, recombinant RINGs, and His-ubiquitin (Xie et al. 2002; Zhang et al. 2007). In the presence of E1, E2, and ubiquitin, PEX2RING, PEX10RING, and PEX12RING all exhibited monoubiquitination activity, whereas no activity was observed with MBP alone or in the absence of E1 or E2 (Figure 1C–E).

Figure 1.

Arabidopsis RING peroxins possess E3 ligase activities.
(A) Sequence alignment of the RING domains of Arabidopsis PEX2, PEX10, and PEX12, which were used for generating PEXRING constructs in this study. Positions of zinc binding Cys and His residues are marked by asterisks. Identical residues are shaded in black whereas similar residues are boxed. (B) Purified recombinant MBP-PEX2RING, MBP-PEX10RING, and MBP-PEX12RING proteins at expected molecular weights of ∼51, 53, and 54 kDa, respectively. Numbers on the left are molecular weight (MW) markers in kDa. (C–E)In vitro ubiquitination assays to test for E3 ligase activity. Assays were carried out in the presence of wheat E1, human UBCH5b, 6xHis tagged ubiquitin, and MBP fused PEX2RING(C), PEX10RING(D), or PEX12RING(E). Reactions were analyzed with immunoblots using anti-His antibodies. Sizes of molecular weight markers are indicated on the left in kDa. Arrowheads point to the RING-Ub conjugates, and asterisks indicate His-Ub.

In parallel, we used another in vitro system with a rabbit E1 and another human E2 (UBCH5c) to test for E3 ligase activity. We also observed monoubiquitination activity of PEX2RING and PEX10RING, and in the case of PEX12RING a possibly enhanced E3 activity was detected (Figure S1A).

The RING motif forms a cross-brace structure in which eight metal ligands, usually cysteine and histidine, coordinate two zinc (Zn) ions (Deshaies and Joazeiro 2009). While PEX2 and PEX10 have all of the eight conserved metal ligands, PEX12 only has five of the eight conserved residues for Zn binding (Figure 1A), which have been shown to bind a single zinc ion in yeast (Koellensperger et al. 2007). Thus, to further test the importance of the RING structure to the E3 activity of PEX2RING and PEX10RING, we used a Zn chelator, N,N,N’,N’-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN), to disrupt the RING domains of PEX2 and PEX10, and tested whether the disrupted RING domains still possess E3 activity using a previously published in vitro E3 assay system that employs a yeast E1 and an Arabidopsis E2 (UBC8) (Stone et al. 2005). The well-characterized E3 ligase COP1-Interacting Protein 8 (CIP8; Hardtke et al. 2002) was used as a positive control. In the presence of E1, UBC8 and bovine ubiquitin, CIP8, PEX2RING, and PEX10RING showed E3 ligase activity (Figure 2A). Depletion of Zn by TPEN treatment abolished ubiquitin ligase activity for the E3s, whereas the lost E3 activity of TPEN-treated samples could be restored by incubation with zinc chloride (Figure 2A). Immunoblotting of the samples with MBP antibodies confirmed that PEX2RING and PEX10RING undergo Zn-dependent autoubiquitination (Figure 2B).

Figure 2.

Effect of zinc (Zn) depletion on the E3 ligase activity of PEX2RING and PEX10RING.
GST-CIP8, MBP-PEX2RING, and MBP-PEX10RING were subject to mock (5% ethanol) and 5 mmol/L N,N,N’,N’-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN) treatment, and TPEN followed by 1 mmol/L ZnCl2, which restored the E3 activity in the presence of yeast E1 and Arabidopsis UBC8. Reactions were analyzed with anti-ubiquitin (A) or anti-MBP (B) antibodies. Arrowhead indicates PEX2/10RING-Ub. Asterisks indicate PEX2/10RING. Low molecular weight (MW) bands cross-reacting to Ub antibodies are mostly E2-dependent ubiquitination products, as they were not observed in the minus E2 samples (Figure S1B, C). Numbers to the left of the gels are MW markers in kD.

Taken together, our data demonstrate that the RING domain of the three Arabidopsis RING peroxins, PEX2, PEX10, and PEX12, contain E3 ubiquitin ligase activity. At least for PEX2 and PEX10, the E3 activities are dependent on the binding to Zn ions.

Identification of ubiquitin receptor family proteins DSK2a and DSK2b as PEX2-interacting proteins

The detection of E3 ligase activity for all three Arabidopsis RING peroxins prompted us to search for their functional partners and potential substrates. We began this effort by trying to identify PEX2-interacting proteins in a Y2H screen, using the RING domain of PEX2 (PEX2RING) as bait against an Arabidopsis seedling cDNA library with the GAL4 Y2H system. Sequencing of positive clones revealed a potential interacting protein, DSK2b (At2g17200).

DSK2b belongs to the ubiquitin-like (UBL) and ubiquitin-associated (UBA) domain containing ubiquitin receptor proteins that mediate recognition of ubiquitinated substrates for ubiquitin-proteasome-based degradation (Fu et al. 2010). These proteins bind to ubiquitinated substrates via the UBA domain and to the proteasome subunits through the UBL domain, thus serving as shuttle factors that couple polyubiquitination to proteasomal degradation (Fatimababy et al. 2010). In Arabidopsis, DSK2b has a tandemly duplicated and highly similar paralog, DSK2a (At2g17190), which shares 82% sequence identity with DSK2b at the protein level. The DSK2 proteins each consist of an N-terminal UBL domain, four chaperonin domains in the middle, and a C-terminal UBA domain (Figure 3A). Given the high sequence similarities between DSK2a and DSK2b, we tested both proteins in a LexA matchmaker Y2H system for interaction with the full-length PEX2 and PEX2RING. Positive interactions, revealed by the LacZ reporter activity (blue color) in the colonies, were observed only between the two DSK2s and PEX2RING, in addition to the positive control (Figure 3B), suggesting that both DSK2a and DSK2b are capable of interacting with PEX2RING. The expression of the fusion proteins in yeast was verified by immunoblotting (Figure S2). The absence of interaction with the full-length PEX2 could be due to mistargeting or misfolding of the proteins, or steric constraints that obscured the RING domain.

Figure 3.

DSK2a and DSK2b are PEX2-interacting proteins.
(A) Schematic showing the domain organization of the DSK2 protein. Grid-lined boxes numbered 1–4 represent the four chaperonin domains. Domains are not drawn to scale. (B) Yeast two-hybrid (Y2H) analysis to test for interaction between DSK2s and PEX2. Yeast strain EGY48/p8opLacz transformed with AD and BD fusion constructs were plated on selection media containing galactose and X-gal. Blue color indicates interaction and white indicates no interaction. Positive control strain used contained LexA-p53 and pB42AD-T antigens, while transformants containing empty vectors served as negative controls. (C)In vitro pull-down assays to confirm the interaction between DSK2 and PEX2RING. Recombinant 6xHis-DSK2 retains MBP-PEX2RING as detected by MBP antibodies. His antibodies were used to confirm the presence of the bait proteins.

We then used in vitro pull-down assays to validate the Y2H interaction results by expressing recombinant DSK2a and DSK2b as 6xHis-tagged fusion proteins and PEX2RING as an MBP fusion protein, and assessed the ability of Ni-NTA-bound His-DSK2 to retain or pull down MBP-PEX2RING in bacterial lysates. Both DSK2a and DSK2b were capable of pulling down MBP-PEX2RING (Figure 3C). We thus conclude that DSK2a and DSK2b do indeed interact with the RING domain of PEX2.

Previous studies on DSK2 showed that the N-terminal UBL domain interacts with the proteasome subunit RPN10 through the ubiquitin interaction motif (UIM)1, and also weakly interacts with the RPN13 subunit via a PRU (pleckstrin-like receptor of ubiquitin) domain (Fatimababy et al. 2010; Lin et al. 2011). The UBA domain associates in high affinity with K48-linked ubiquitin, and is required for interaction with ubiquitinated proteins (Lin et al. 2011). To delineate which domain is responsible for interacting with the RING peroxins, we made truncated constructs with UBL, chaperonin 1 and 2, chaperonin 3 and 4, and UBA domains respectively deleted, and tested the ability of the truncated proteins to interact with PEX2RING in Y2H. Our data showed that the chaperonin domains 3 and 4 play a strong role in the interaction with PEX2RING, and chaperonin domains 1 and 2 and the UBL domain are also involved in the interaction, albeit to lesser degrees (Figure 4A).

Figure 4.

Yeast two-hybrid (Y2H) analyses to dissect DSK2 domains responsible for interaction with PEX2RING and to test for DSK2's interaction with other Arabidopsis RING proteins.
(A) Y2H assays to assess which domain in DSK2 is responsible for its interaction with PEX2RING. Schematics show the deletion constructs used in the assays. (B) Y2H assays to test for interaction between DSK2s and other RING proteins. The full-length RING peroxins PEX10 and PEX12 and their RING domain constructs (PEX10RING and PEX12RING), and two representative proteins from two other types of RING domains (ARI8 from HCb type and At2g44330 from H2 type) were tested. Positive control strain used contained LexA-p53 and pB42AD-T antigens, while transformants containing empty BD and AD vectors served as the negative control.

RING domain of PEX12 also interacts with DSK2a and DSK2b

RING domain proteins have been implicated in a diverse range of biological processes, and make up one of the largest protein families in the Arabidopsis proteome (Budhidarmo et al. 2012). A hallmark of the RING domain is the presence of eight Cys/His residues that coordinate two Zn ions. On the basis of the Cys/His residues present in the RING domain and the number of amino acids between them, the 477 Arabidopsis RING proteins are classified into two major classes, H2 and HC, of which HC is further subdivided into HCa and HCb (Stone et al. 2005). The RING peroxins PEX2, PEX10, and PEX12 belong to the HCa class of RING proteins.

To determine whether DSK2s also interact with the other two RING peroxins and promiscuously with other types of RING proteins, we tested the interaction between the DSK2s (bait) and other RING proteins (prey) in Y2H assays. In addition to PEX10 and PEX12, ARI8 (At1g65430, an HCb) and the protein encoded by At2g44330 (H2) were also included. Considering that PEX10 and PEX12 are integral membrane proteins and that only PEX2RING interacted with DSK2s, we also used the RING domains of the two peroxins in the assay. Because ARI8 has two RING domains in the N-terminal region of the protein (Mladek et al. 2003), we also made a truncated construct that expressed only the RING domains (ARI8RING). Our results demonstrated that the DSK2s interacted strongly with PEX12RING, but very weakly with PEX10RING and not at all with the other two RING proteins tested (Figure 4B; Figure S3).

DSK2a and DSK2b are localized in the cytosol and nucleus and are ubiquitously expressed in plants

To analyze the subcellular localization of DSK2a and DSK2b, we made YFP-DSK2 fusion constructs driven by the constitutive 35S promoter and expressed them transiently in Nicotiana tabacum (tobacco). Confocal microscopic analysis of leaf epidermal cells revealed that both DSK2a and DSK2b are localized to the cytosol and nucleus (Figure 5A). We also made transgenic plants co-expressing YFP-DSK2 and the peroxisome marker CFP-PTS1. Confocal images taken from transgenic plants showed that, similar to results from the transient expression, YFP-DSK2 proteins are not targeted to peroxisomes but instead localize to the cytosol and the nucleus (Figure S4A).

Figure 5.

Subcellular localization of the DSK2 proteins and expression analysis of the DSK2 genes.
(A) Confocal images taken from tobacco leaf epidermal cells transiently expressing YFP-DSK2 fusion proteins. DAPI was used for nuclear staining. Merged images show the co-localization of some YFP-DSK2 proteins with the nucleus. Scale bar is 10 μm. (B) Reverse transcription polymerase chain reaction analysis of total RNA extracted from wild-type 10 d old seedlings. I, inflorescence; F, flowers; S, stems; R, rosette leaves; and C, cauline leaves. The UBQ10 transcript was used as a loading control.

We initially searched the Arabidopsis electronic fluorescent pictographic browser ( for information on the expression of the DSK2 genes from microarray datasets, but discovered that the probe sets do not distinguish between the highly similar DSK2a and DSK2b genes (Winter et al. 2007). To examine the expression pattern of the two genes, we conducted reverse transcription polymerase chain reaction (RT–PCR) to analyze their transcript levels in various plant tissues, including 10 d old seedlings, inflorescence, flowers, stems, and rosette and cauline leaves. Although both DSK2 genes are expressed in all tissues, DSK2b is expressed at a much higher level than DSK2a in most tissues (Figure 5B).

DSK2 amiRNA lines do not have obvious defects in plant growth or peroxisomal import and function

To evaluate the role of DSK2a and DSK2b in Arabidopsis development and peroxisome function, we obtained T-DNA insertion mutants of DSK2a and DSK2b genes, but none of the homozygous mutants showed altered transcript levels of the two genes (Figure S4B, C). DSK2a and DSK2b are tandem duplicates on chromosome 2, thus making the generation of a double mutant a difficult task. To this end, we made amiRNA constructs that target two conserved regions of the transcripts (see Materials and Methods) and transformed Col-0 plants with the constructs. Transgenic lines with significantly reduced expression of both genes were identified (Figure 6A). However, these lines did not show any noticeable growth or developmental abnormalities.

Figure 6.

Identification and phenotypic analysis of DSK2 amiRNA lines.
(A) Reverse transcription polymerase chain reaction analysis of the DSK2a and DSK2b transcripts in the amiRNA lines. UBQ10 was used as the loading control. (B) Hypocotyl lengths of amiRNA seedlings grown in the dark for 7 d in the presence or absence of 1% sucrose. Error bars indicate standard deviations of n > 25. Student's t-test analyses did not reveal any statistically significant differences compared to the wild-type, except for pex14 plants (indicated by an asterisk). (C) Relative root length (on indole-3-butyric acid (IBA) vs without IBA) of amiRNA lines grown on increasing concentrations of IBA in light. Error bars indicate standard deviations of n > 25. Student's t-test analyses did not reveal any statistically significant differences for the amiRNA lines compared to the wild-type. (D) Confocal images from leaf epidermal cells of 14 d old amiRNA lines expressing the peroxisome marker CFP-PTS1. Scale bar is 10 μm. A1, A4, B3, and B4 are independent DSK2 amiRNA lines.

Given that DSK2a and DSK2b interact with two of the RING peroxins, PEX2 and PEX12, we used physiological assays to ascertain if the amiRNA lines displayed any defects in peroxisome metabolism. Peroxisome-based β-oxidation of stored lipids during seed germination provides a carbon source and energy for the seedling; thus, mutants deficient in many peroxisomal proteins are severely compromised in their ability to grow on media lacking exogenous sucrose (Baker et al. 2006). To test the efficacy of β-oxidation, we compared the hypocotyl growth of dark-grown seedlings germinated on Murashige and Skoog (MS) plates with or without 1% sucrose (Figure 6B). Peroxisomes also carry out the conversion of the protoauxin IBA to the active auxin indole-3-acetic acid (IAA), causing root inhibition in plants (Zolman et al. 2000). Thus, we tested the capacity of the amiRNA lines to effectively metabolize IBA by measuring root lengths of seedling germinated on increasing concentrations of IBA (Figure 6C). In both assays, DSK2 amiRNA lines showed no difference from the wild-type plants, whereas the peroxisome biogenesis factor mutant pex14 displayed sugar dependence and IBA resistance (Figure 6B, C), leading us to conclude that peroxisome metabolism is unaffected in the absence of DSK2 function.

Because PEX2 and PEX12 are important constituents of the peroxisome matrix protein import machinery, we checked if DSK2 silencing affected peroxisomal protein import. A peroxisome marker protein consisting of cyan fluorescent protein (CFP) fused to the tripeptide Ser-Lys-Leu, a type 1 peroxisome targeting signal (PTS1), was introduced into the amiRNA plants. Confocal microscopy was used to assess protein import in T2 transgenic plants. No discernible defects were observed in the amiRNA plants in the peroxisomal targeting of CFP-PTS1, peroxisome morphology, or peroxisome abundance (Figure 6D). These data suggest that DSK2s are not directly involved in regulating peroxisome functions, or that they may be acting redundantly with other ubiquitin receptors. It is also possible that some yet-unknown compensatory mechanisms are invoked to alleviate the effect of DSK2 silencing in Arabidopsis.


In this study, we have shown that like their yeast counterparts, all three Arabidopsis RING peroxins can function as E3 ubiquitin ligases. One notable difference is that in yeast, PEX2 and PEX10 have been reported to be polyubiquitin ligases (Platta et al. 2009), but we only found monoubiquitination activity for the Arabidopsis proteins. Due to the insoluble nature of the full-length proteins, we used the RING domain to test for activity. It is thus possible that, in vivo, the full-length proteins do possess polyubiquitination activity, and may even act synergistically as multiprotein complexes to exert their functions, as recently shown for the yeast RING peroxins (El Magraoui et al. 2012). Furthermore, in vitro assays have their limitations, and we thus cannot rule out the possibility that specific stimuli or accessory proteins exist within the cell that are required to promote catalytic activity. Moreover, some E2–E3 combinations are specifically linked with monoubiquitination, as seen in case of the DNA repair pathway, where Rad6-Rad18 monoubiquitination of the proliferating cell nuclear antigen activates recruitment of DNA polymerases (Hibbert et al. 2011). Recycling of most of the characterized PTS receptors in yeast appears to be dependent on monoubiquitination (Kiel et al. 2005; Kragt et al. 2005; Carvalho et al. 2007; Platta et al. 2007; Williams et al. 2008; Hensel et al. 2011). Thus, Arabidopsis RING peroxins may also play a role in PEX5 recycling by targeting PEX5 for monoubiquitination, although other targets may also exist (Hu et al. 2012). Finally, monoubiquitination plays a role in various cellular processes such as the trafficking of ubiquitinated endosomal receptors, the modulation of protein activity, transcriptional regulation, and the facilitation of protein interactions (Hicke 2001). It has also been reported that the activity of proteins with ubiquitin binding domains (UBDs) can be regulated by monoubiquitination that results in inhibition of their binding to ubiquitinated substrates (Hoeller et al. 2006). Considering that DSK2s also harbor UBD, namely, C-terminal UBA, and associate with PEX2 and PEX12, they may undergo monoubiquitination. In this scenario, the monoubiquitination of DSK2s would prevent them from recognizing ubiquitinated substrate proteins like PEX5, and would thus increase target protein stability in the cell.

Despite having an incomplete RING domain, PEX12 also exhibits E3 ligase activity. The conserved residues that are changed in PEX12RING are at positions 3, 4, and 8 in the C3HC4 RING domain, namely, Cys->Ser, His->Phe and Cys->Thr (Figure 1A). These substitutions make this domain look more like that of the U-box proteins, a new class of E3 ligases derived from RING proteins. U-box E3 ligases lack the conserved Cys/His and instead use charged and polar residues (Ser, Thr, Asp, Glu) to maintain the structural integrity of the protein (Aravind and Koonin 2000; Ohi et al. 2003). Thus, PEX12RING appears to have features of both the RING and U-box E3 ligases. It is well known that E3s pair with a very narrow set of E2s to carry out substrate-specific ubiquitination (Christensen and Klevit 2009; Ye and Rape 2009). For example, the tomato pathogen Pseudomonas syringae type III effector AvrPtoB, a U-box protein, exhibits E3 ligase activity in the presence of UbcH5a and UbcH5c but not UbcH5b (Abramovitch et al. 2006). Several mammalian U-box ligases also exhibit a preference for UbcH5c (Hatakeyama et al. 2001). This suggests that the choice of E2 is critical in determining the specificity of the E3 assays. In agreement with this, PEX12 showed enhanced activity when UbcH5c was present in the E3 assay in our study. Moreover, UbcH5c is one of the three members of the human E2D family that ubiquitinate PEX5 for its recycling (Grou et al. 2008), further supporting our results.

We also report the identification of the tandemly-duplicated UBL-UBA proteins DSK2a and DSK2b as specific interactors of PEX2RING and PEX12RING. Our expression analysis revealed that DSK2a has much lower expression levels than DSK2b in plants, thus explaining why we did not find DSK2a in our initial Y2H screen. Arabidopsis DSK2s and their homologs in other species have been characterized as molecular adaptors, sometimes called E4, that regulate the relay of ubiquitinated substrate proteins to the proteasome for degradation (Fu et al. 2010). Although the interactions of some of the UBA-UBLs with ubiquitin and proteasomal subunits have been extensively studied, the physiological significance of these interactions remains undefined. In yeast, dsk2 null mutants are viable and only exhibit subtle phenotypes such as increased cellular polyubiquitination (Biggins et al. 1996; Funakoshi et al. 2002). Arabidopsis has at least eight UBA-UBL family proteins, including four Rad23s, two DSK2s, DDI1, and NUB1 (Farmer et al. 2010; Fu et al. 2010). Several lines of evidence suggest that most of these proteins act in a functionally redundant manner. RNAi lines that target DSK2s in Arabidopsis are no different from wild-type plants in appearance (Lin et al. 2011), and neither were our amiRNA lines generated in this study. Mutants of another ubiquitin receptor protein, RPN10 (a component of the proteasome), showed elevated levels of UBA-UBL proteins including DSK2s, suggesting that plants invoke compensatory mechanisms to counter the perturbed ubiquitin receptor levels. This may also account for our failure to find any peroxisome-related phenotypes in the DSK2 amiRNA lines.

In summary, we provide evidence that the Arabidopsis RING peroxins PEX2, PEX10, and PEX12 have monoubiquitin ligase activity, and that the ubiquitin receptor proteins DSK2a and DSK2b specifically associate with PEX2 and PEX12. We speculate that, together with ubiquitin receptor proteins such as DSK2s, the RING peroxins form a peroxisome-based recycling/degradation system in plants. Identification of the target protein(s) of this machinery would be instrumental to defining the role(s) and cellular consequences of such a surveillance system.

Materials and Methods

Sequence alignment

PEX2, PEX10, and PEX12 protein sequences were downloaded from The Arabidopsis Information Resource (TAIR) at RING domains were identified using SMART (; Letunic et al. 2012). Sequence alignment was performed with ClustalW2 software (; Larkin et al. 2007) and rendered in graphical format using ESPript software ver. 2.2 (Gouet et al. 1999).

Protein expression and purification

The RING domains of PEX2 (amino acids (a.a.) 248–333), PEX10 (a.a. 287–381), and PEX12 (a.a. 296–393) were amplified with the primers listed in Table S1. The amplified PCR products were digested with EcoRI and XbaI and cloned into the pMAL-c4x expression vector (New England Biolabs, Ipswich, MA, USA) to generate constructs encoding MBP-tagged proteins. Full-length coding sequences of DSK2a and DSK2b were cut with NheI and EcoRI (DSK2a) or NcoI and BamHI (DSK2b) for cloning into the pET28a plasmid (EMD Millipore, Billerica, MA, USA) to generate 6xHis-tagged proteins. Protein expression constructs were transformed into Escherichia coli strain BL21-DE3 (Stratagene, Clara, CA, USA). At optical density (OD) 0.6, bacterial cultures harboring different constructs were induced with 0.5 mmol/L isopropyl thiob-d-galactoside, incubated at 37 °C for 3 h and harvested by centrifugation. Cell pellet was resuspended in column buffer (200 mmol/L NaCl, 20 mmol/L Tris-HCl, pH 7.4, 1 mmol/L ethylenediamine tetraacetic acid, 1 mmol/L phenylmethylsulfonyl fluoride (PMSF), 1 mmol/L dithiothreitol (DTT)) for the MBP-tagged proteins and lysis buffer (50 mmol/L NaH2PO4, pH 7.5, 300 mmol/L NaCl, 10 mmol/L imidazole, 1 mmol/L PMSF) for His-tagged proteins. Cells were disrupted by sonication in an ice-water bath (4 °C), and lysed cells were centrifuged at 7,350 g for 10 min. Cleared lysates of MBP-tagged protein were applied to an amylose resin column and purified according to manufacturer's instructions (New England Biolabs).

In vitro ubiquitination assays

For the assays presented in Figure 1C–E, crude extracts containing recombinant wheat E1, human UbcH5b (E2; ∼40 ng), purified MBP-PEX2RING, MBP-PEX10RING or MBP-PEX12RING (E3; ∼1 μg), and purified His-ubiquitin (∼2 μg) were used for E3 ubiquitin ligase activity assays as described previously (Xie et al. 2002; Zhang et al. 2007). Reactions were stopped by adding 4X sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) buffer and samples were separated on 12% SDS-PAGE gels followed by western blotting analysis using anti-His antibody (Santa Cruz Biotechnology, CA, USA).

The assays presented in Figure 2 and Figure S1 were performed according to Stone et al. (2005). For the zinc-chelating assays, RING proteins bound to the amylose beads were incubated in column buffer containing either mock (Ethanol) or 5 mmol/L TPEN (Sigma-Aldrich, St Louis, MO, USA) for 16 h at 4 °C on a rotary shaker. Solutions were changed every 4 h, after which beads were washed twice with column buffer. TPEN-treated fraction was split into two aliquots, one of which was further incubated with 1 mmol/L ZnCl2 for 4 h at 4 °C with three solution changes followed by three washes with column buffer. The bead-bound RING proteins thus treated (Mock, TPEN, or TPEN + ZnCl2) were subsequently used in ubiquitination assays. Ubiquitination assays were done using 50 ng of yeast E1 (Boston Biochem, Cambridge, MA, USA), 250 ng of AtUBC8, 500 ng of bead-bound MBP-RING protein, and 2 μg of bovine ubiquitin (Sigma-Aldrich) incubated in reaction buffer comprised of 50 mmol/L Tris-HCl, pH 7.5, 10 mmol/L MgCl2, 0.05 mmol/L ZnCl2, 1 mmol/L adenosine triphosphate, 0.2 mmol/L DTT, 10 mmol/L phosphocreatine and 0.1 unit of creatine kinase (Sigma-Aldrich) for 2 h at 30 °C. Reactions were stopped by adding 5X SDS-PAGE buffer, split, and run on 10% NuPAGE gels (Invitrogen, Grand Island, NY, USA) followed by western blotting analysis using either ubiquitin (Sigma-Aldrich) or MBP (New England Biolabs) antibodies. E3 assays shown in Figure S1A were performed in the same way, except that rabbit E1 and human UbcH5c were used instead of yeast E1 and AtUBC8.

Yeast two-hybrid (Y2H) assays

An Arabidopsis cDNA library made from seedlings was screened for proteins interacting with PEX2, using the GAL4 Y2H system with PEX2RING as bait. The Matchmaker LexA system (Clontech, Mountain View, CA, USA) was used to perform further yeast two-hybrid (Y2H) assays to confirm the interactions. Coding sequences for full-length PEX2, PEX10, PEX12, ARI8 (At1g65430; HCb), and At2g44330 (H2), and deletions comprising only the RING domain, were amplified by PCR and cloned into the pGILDA bait vector to generate LexA DNA binding domain fusions. Full-length coding sequences for DSK2a and DSK2b and deletions thereof were PCR amplified and cloned into a modified pB42AD plasmid to generate prey constructs. Primers used for cloning are listed in Table S1. To test the interaction between DSK2 and the RING domain proteins, the bait and prey constructs were co-transformed into yeast (Saccharomyces cerevisae) strain EGY48 (p8opLacZ), using the transformation protocol described previously by Gietz and Woods (2002). Yeast strains transformed with empty bait/prey vectors were used as negative controls, and the yeast strain transformed with pLexA-53 and pB42AD-T plasmids was used a positive control. Transformants were selected on SD-glucose (BD Biosciences, San Jose, CA, USA) media supplemented with -Ura/-His/-Trp synthetic dropout solution. Transformants were grown overnight in SD-glucose -Ura/-His/-Trp liquid medium, centrifuged, washed twice with distilled water, and plated on SD-galactose/raffinose -Ura/-His/-Trp-inducing media containing 80 μg/mL of X-gal. Plates were incubated at 30 °C and imaged after 48 h. LexA (Invitrogen) and HA-tag antibodies (Aves Labs, Tigard, OR, USA) were used to detect BD and AD-fusion proteins.

In vitro pull-down assays

Cleared lysates of His-tagged DSK2 proteins were incubated with Ni-NTA resin (Qiagen, Valencia, CA, USA) for 1 h at 4 °C. The resin was washed three times with lysis buffer and incubated with lysate from MBP-PEX2RING for 2 h at 4 °C with gentle agitation. Ni-NTA resin was recovered by low speed centrifugation, and was further washed with lysis buffer at least three times. Bead-bound proteins were then eluted with lysis buffer supplemented with 250 mmol/L imidazole. Elutes were subjected to immunoblotting using His (Cell Signaling, Danvers, MA, USA) or MBP (New England Biolabs) antibodies to determine interactions.

Subcellular protein localization

The open reading frames of DSK2a and DSK2b were PCR-amplified with Gateway primers containing the attB1 and attB2 sequences. The PCR product was recombined into the Gateway entry vector pDONR207 (Invitrogen) using BP clonase. The pDONR clones were transferred to binary destination vector pEARLEY104 via LR clonase-based recombination, resulting in constructs encoding YFP-DSK2 fusions. Agrobacterium tumefaciens strain GV3101 (pMP90) transformed with the constructs of interest was grown overnight at 28 °C, washed and resuspended in water to OD600= 0.1. Tobacco plants were infiltrated with the bacterial suspension using a needleless syringe and kept in regular growth conditions for 2 d.

An inverted Zeiss LSM 510 Meta confocal microscope (Carl Zeiss, Thornwood, NY, USA) was used for all fluorescent protein imaging. To label the nuclei, the fluorescent dye 4′,6′-diamidino-2-phenylindole dihydrochloride (DAPI; Sigma-Aldrich) was diluted to 10 μg/mL in phosphate buffer at pH 7.2, and the solution was infiltrated into the tobacco leaves 1 h prior to imaging. To observe subcellular localization, we used a 405 nm diode and a 514 nm argon laser to excite DAPI and YFP, respectively. The fluorescent emission from DAPI and YFP was acquired through the 420–480 nm and 520–555 nm band-pass filters, respectively. All images were obtained from a single focal plane. For imaging CFP, we used the 458 nm argon laser for excitation and the 465–510 nm emission filter to detect the CFP fluorescence.

Generation of amiRNA lines

The WMD3-Web MicroRNA Designer ( tool was used to find the best target amiRNA for DSK2a and DSK2b. Two of the targets, designated amiRNA-a (5′-TAGTCGTTCTACAGCTGCGTT-3′) and amiRNA-b (5′-TATCATTTCACGCATACGCTC-3′), were selected and amplified from the miR319a backbone using overlapping PCR as described in the amiRNA cloning protocol ( The amiRNA precursors were then digested with KpnI and XbaI and cloned into binary vector pCHF3. The constructs were transformed into A. tumefaciens strain GV3101, which was then transformed into Col-0 and CFP-PTS1 Arabidopsis plants by the floral-dip method (Clough and Bent 1998). Transgenic plants were selected on 1/2 Murashige and Skoog (MS) supplemented with 0.5% sucrose and 50 μg/mL kanamycin. T1 antibiotic-resistant plants were screened with RT–PCR to identify individual plants in which the expression of DSK2a and DSK2b was reduced. Segregation of the transgenes in subsequent generations was assessed with resistance to kanamycin to identify homozygous lines, which were subsequently used for physiological assays.


Total RNA was isolated from 3-w-old plants using the RNeasy Plant Mini Kit (Qiagen) as per the manufacturer's instructions. RNA (1 μg) was reverse-transcribed with the Omniscript RT Kit (Qiagen) using oligodT primers. Subsequently, 50 ng of cDNA was used in PCR (Promega, Madison, WI, USA) amplification using gene-specific primers listed in Table S1. PCR conditions used were as follows: 95 °C for 2 min, 30 cycles of 95 °C for 30 s, 57 °C for 30 s, 72 °C for 30 s, and a final extension of 72 °C for 10 min.

Physiological assays

To test for sucrose dependence, seeds from wild-type and amiRNA lines were surface-sterilized and plated on 1/2 MS growth medium solidified with 0.6% phytagar, in the presence or absence of 1% sucrose. Plates were wrapped in foil, stratified for 2 d at 4 °C, and imaged after 7 d of growth in the dark at 22 °C.

For IBA response assays, seeds were sown on 1/2 MS medium supplemented with 0.5% sucrose and various concentrations of IBA (Sigma-Aldrich). Following 2 d of stratification at 4 °C, the plates were placed vertically in a growth chamber (Percival, Promega, Madison, WI, USA) with continuous light, covered with a mesh, and scanned after 7 d of growth.

ImageJ software ( was used to measure both the hypocotyl length of the etiolated seedlings in the case of the sucrose dependence assay, and root length for the IBA response assays. A previously-identified peroxisome import mutant, pex14 (SALK_007441), served as the positive control in both assays.

(Co-Editor: Giovanna Serino)


The authors would like to thank the Arabidopsis Biological Resource Center (ABRC) for providing seeds of the DSK2 T-DNA insertion mutants, and Jilian Fan for genotyping the mutants, Judy Callis (University of California, Davis) for the UBC8 and CIP8 constructs, Sheng Yang He (Michigan State University) for the modified pB42AD plasmid, and Detlef Weigel (Max Planck Institute for Developmental Biology, Tübingen, Germany) for sharing the amiRNA backbone pRS300 vector. This work was supported by grants from the National Science Foundation Arabidopsis 2010 program (MCB 0618335) and the Chemical Sciences, Geosciences and Biosciences Division, Office of Basic Energy Sciences, Office of Science, US Department of Energy (DE-FG02–91ER20021) to J. H.