At least two of the genes predicted to encode type II PI4K (phosphoinositide 4-kinase) in Arabidopsis thaliana (thale cress), namely AtPI4Kγ4 and AtPI4Kγ7, encode enzymes with catalytic properties similar to those of members of the PIKK (phosphoinositide kinase-related kinase) family. AtPI4Kγ4 and AtPI4Kγ7 undergo autophosphorylation and phosphorylate serine/threonine residues of protein substrates, but have no detectable lipid kinase activity. AtPI4Kγ4 and AtPI4Kγ7 are members of a subset of five putative AtPI4Ks that contain N-terminal UBL (ubiquitin-like) domains. In vitro analysis of AtPI4Kγ4 indicates that it interacts directly with, and phosphorylates, two proteins involved in the ubiquitin–proteasome system, namely UFD1 (ubiquitin fusion degradation 1) and RPN10 (regulatory particle non-ATPase 10). On the basis of the present results, we propose that AtPI4Kγ4 and AtPI4Kγ7 should be designated UbDKγ4 and UbDKγ7 (ubiquitin-like domain kinases γ4 and γ7). These UBL-domain-containing AtPI4Ks correspond to a new PIKK subfamily of protein kinases. Furthermore, UFD1 and RPN10 phosphorylation represents an additional mechanism by which their function can be regulated.

INTRODUCTION

Enzymes that contain the PI3/4K (phosphoinositide 3/4-kinase domain [EMBL–EMI Database (http://www.ebi.ac.uk/interpro/) InterPro IPR000403] include (i) PIKs (phosphoinositide kinases), (ii) protein kinases and (iii) kinases with dual activity that can use phosphoinositides and proteins as substrates. This diverse group of enzymes is involved in many different cellular functions, including lipid- and protein-mediated signalling [14].

The genome of the model plant Arabidopsis thaliana (thale cress) contains at least 18 distinct loci which encode predicted PI3/4K-domain-containing proteins. A total of 12 of them are predicted to encode AtPI4Ks (A. thaliana phosphoinositide 4-kinases), four belonging to the type III group (AtPI4Kα1, AtPI4Kα2, AtPI4Kβ1 and AtPI4Kβ2) and eight belonging to the type II group (AtPI4Kγ1–AtPI4Kγ8) [5]. Many of the type III AtPI4Ks have been characterized [69]; however, little is known about the type II PI4Ks in plants.

The type II PI4Ks in yeast and mammals have been cloned and characterized. In Saccharomyces cerevisiae (baker's yeast) this enzyme, named LSB6 (Las binding protein 6), accounts for a small percentage of the PI4K activity [10,11], while in humans (Homo sapiens, Hs), the type II PI4Ks HsPI4KIIα and HsPI4KIIβ contribute significantly to PtdIns4P production in both plasma membrane and endomembranes (for a review, see [2]).

Five of the predicted Arabidopsis type II AtPI4Ks (γ3, γ4, γ5, γ6 and γ7) contain one or two N-terminal UBL (ubiquitin-like) domains. None of the other known PI3/4K-domain-containing proteins have a predicted UBL domain. UBL domains consist of 70–100 amino acids and have significant identity with the primary sequence and structure of ubiquitin. UBL-domain proteins are not like ubiquitin or the type I ubiquitin-like proteins such as SUMO (small ubiquitin-related modifier) in that UBL proteins do not covalently bind to target proteins.

UBL domains are usually found at the N-terminus and are associated with a very diverse repertoire of other protein domains [1214]. In many UBL proteins the UBL domain promotes protein–protein interaction with components of the 26 S proteasome [12,1518]. Because of the presence of the UBL domain in what had been predicted originally to be type II PI4Ks [5], we asked whether these purported Arabidopsis type II PI4Ks might have other functions.

Here we show that a subset of an Arabidopsis gene family that is predicted to encode type II PI4Ks and that contains UBL domains encodes active protein kinases, specifically AtPI4Kγ4 and AtPI4Kγ7. Protein–protein interaction studies reveal that AtPI4Kγ4 interacts directly with RPN10 (regulatory particle non-ATPase 10), a non-ATPase component of the 19 S regulatory particle of the 26S proteasome. Furthermore, AtPI4Kγ4 interacts directly with ubiquitin fusion degradation 1 (UFD1). Both RPN10 and UFD1 bind to some form of ubiquitin and are known to facilitate substrate delivery to the 26 S proteasome. AtPI4Kγ4 is a serine/threonine kinase that phosphorylates both RPN10 and UFD1 in vitro. On the basis of the present findings we propose to re-annotate AtPI4Kγ4 and AtPI4Kγ7 as a new subfamily of the PIKK (PIK-related kinase) family of protein kinases and rename them UbDKγ4 and UbDKγ7 (ubiquitin-like domain kinase γ4 and γ7) respectively.

EXPERIMENTAL

cDNA cloning and construction of expression vectors

Full-length AtPI4Kγ1 (At2g40850), AtPI4Kγ4 (At2g46500), and AtPI4Kγ7 (At2g03890) coding regions were amplified via RT-PCR (reverse-transcription PCR) from the total RNA of 4-day-old Arabidopsis cells growing in suspension culture. The primers used were designed to contain restriction-enzyme sites (see Supplementary Table S1 at http://www.BiochemJ.org/bj/409/bj4090117add.htm) to facilitate further cloning. AtPI4Kγ1-, AtPI4Kγ4- and AtPI4Kγ7-coding regions were cloned into the plasmid pGEM®-T Easy (Promega, Madison, WI, U.S.A.) according to manufacturer's protocol and the clones were confirmed by sequencing. Eschercihia coli expression cassettes were made in pET-41 vectors (Novagen, San Diego, CA, U.S.A.) using restriction enzymes as indicated in Supplementary Table S1. The AtPI4Kγ4 sequence encoding the full-length protein and the truncations ΔUBL1, ΔUBL1/UBL2, ΔN and ΔC were PCR-amplified, cloned into the pENTR/SD/D-TOPO entry vector (Invitrogen, Carlsbad, CA, U.S.A.) and verified by sequencing. The resulting entry clones were recombined with the E. coli expression vectors pDEST15 and pDEST17 (Invitrogen) for the production of N-terminal GST (glutathione transferase)-tagged or His6 (hexahistidine)-tagged fusion proteins respectively. The AtPI4Kγ4 K284A mutant was generated by PCR using the Quik Change® site-directed mutagenesis kit (Stratagene, La Jolla, CA, U.S.A.) with the following oligonucleotide primers:

5′-GTGGGTGTGTTTGCGCCAATAGATGAGGAACCAATGGC-3′

5′-GCCATTGGTTCCTCATCTATTGGCGCAAACACACCCAC-3′.

The full-length cDNA clones of the putative AtPI4Kγ-interacting proteins listed in the Table 2 (below) and for CDC48C (locus At3g01610, cDNA clone U24535) were ordered from the ABRC (Arabidopsis Biological Resource Center, Ohio State University, Columbus, OH, U.S.A.) [19] with the exception of Arabidopsis UFD1. UFD1 cDNA was amplified via RT-PCR from total RNA of Arabidopsis leaves (see Supplementary Table S1 for primer sequences) and cloned into the pENTR/SD/D-TOPO entry vector (Invitrogen). The cDNA from the ABRC clones U09559, U22122, U60768 and U09430, which were supplied on pUNI51 vectors, were subcloned via PCR into the pENTR/SD/D-TOPO entry vector (see Supplementary Table S1 for primer sequences). All entry clones were recombined with the E. coli expression vectors pDEST15 and pDEST17 (Invitrogen) for production of N-terminal GST- or His6-tagged fusion proteins respectively.

Recombinant protein production and purification

The various expression cassettes were transformed into E. coli expression strain BL21 (DE3) pLysS (Invitrogen). Production and affinity purification of recombinant GST-tagged proteins was performed as described by Perera et al. [20], using Glutathione–Sepharose™ 4B (Amersham Biosciences; part of GE Healthcare, Piscataway, NJ, U.S.A.). For some studies, proteins were eluted with elution buffer containing reduced glutathione according to the manufacturer's protocol. The His6-tagged protein purification was performed using HIS-select™ Nickel Affinity Gel (Sigma, St. Louis, MO, U.S.A.) according to the manufacturer's protocol. Protein concentrations were determined using the Bradford method (Bio-Rad, Hercules, CA, U.S.A.) with BSA as a standard.

Protein immunoblotting

Tagged proteins were detected by immunobloting using monoclonal anti-Penta·His™ (Qiagen G.m.b.H., Hilden, Germany) antibodies at a dilution of 1:1000 for 1 h, followed by treatment with horseradish-peroxidase-conjugated anti-mouse antibody as the secondary antibody at a dilution of 1:20000 for 1 h. Immunoreactivity was visualized by incubating the blot with SuperSignal West Pico Chemiluminescence Substrate (Pierce, Rockford, IL, U.S.A.) and subsequent exposure to X-ray film. After chemiluminescence detection, total protein was visualized by staining the blots with Amido Black (Sigma).

In vitro phosphorylation assays

Recombinant proteins (immobilized on affinity beads or eluted) were assayed for lipid and protein kinase activity and for autophosphorylation. Protein kinase activity and autophosphorylation were assayed in buffer containing 50 mM Tris/HCl (pH 7.5), 10 mM MgCl2, 1 mM EGTA, 100 μM ATP (5 μCi of [γ-32P]ATP/reaction) in the absence of substrate (autophosphorylation assay) or in the presence of type III-S histone (Sigma), MBP (myelin basic protein) (Sigma) or other proteins as indicated as phosphate-acceptor substrates. The optimal Mg2+ concentration was analysed, and phosphorylation efficiency was indistinguishable (results not shown) at concentrations between 2.5 and 10 mM MgCl2. Typically, reactions were carried out in 40–60 μl (final volume) for 10–20 min at room temperature (23 °C) with occasional agitation and were terminated by adding 4×SDS sample buffer. Reaction mixtures were resolved by SDS/PAGE, followed by autoradiography to visualize phosphate incorporation.

PIK activity was assayed in 50 μl reaction volumes containing 50 mM Tris/HCl (pH 7.5), 10 mM MgCl2, 1 mM EGTA, 0.1% Triton X-100, 100 μM ATP (9 μCi of [γ-32P]ATP per reaction) in the presence of 0.25 mg/ml PtdIns or 0.3 mg/ml Type I Folch fraction I (Sigma) from bovine brain extract. Stock PtdIns and Type I Folch fraction I (5 mg/ml) were solubilized in 1% Triton X-100. The reaction mixtures were incubated at room temperature for 20 min with intermittent shaking. The reactions were stopped with 1.5 ml of ice-cold chloroform/methanol (1:2, v/v) and kept at 4 °C until the lipids were extracted, then analysed as described previously [6,7].

Protein pull-down assays

GST-tagged proteins immobilized in glutathione–Sepharose beads were incubated with either Arabidopsis cell extracts or E. coli-expressed recombinant proteins containing a His6 tag. Arabidopsis cell extracts were prepared from 5-day-old cells growing in suspension cultures. The cells were harvested by filtration and immediately homogenized in cold TEEM buffer (50 mM Tris/HCl, pH 7.5, 2 mM EGTA, 2 mM EDTA and 1 mM MgCl2) supplemented with 200 mM sucrose, 1 mM PMSF and 10 μg/ml leupeptin. Homogenization was carried out in a glass Dounce homogenizer with 1% polyvinylpolypyrrolidone to facilitate grinding. The crude extract was clarified by centrifugation at 5000 g for 5 min at 4 °C. The supernatant was fractionated further (40 000 g, for 60 min at 4 °C) to yield microsomal and soluble protein fractions. The microsomal pellet was solubilized in non-supplemented TEEM buffer. Microsomal and soluble fractions were pretreated by incubating with purified GST immobilized on glutathione–Sepharose beads in a 10:1 ratio to remove non-specific binding proteins. Precleared fractions were incubated with immobilized GST–AtPI4Kγ7 and GST–AtPI4Kγ4 in TEEM buffer for 2 h at 4 °C. After extensive washing, the bound proteins were separated by SDS/PAGE. Selected bands revealed by Coomassie Blue staining were excised from the gel and subjected to LC–MS/MS (liquid chromatography–tandem MS) analysis for identification.

Pull-downs using recombinant proteins were performed by incubating GST-tagged proteins immobilized in glutathione–Sepharose beads with purified His6-tagged proteins in the presence of binding buffer (50 mM Tris/HCl, pH 7.5, 100 mM KCl and 0.05% Triton X-100). After extensive washing with binding buffer, bound proteins were added directly into 4×SDS/PAGE sample buffer, boiled for 5 min, separated by SDS/PAGE and subjected to immunoblot analysis. GST-tagged proteins were visualized directly by Amido Black staining of the blot. Recovered His6-tagged proteins were detected by immunoblot with monoclonal Penta·His™ antibody.

Protein identification by LC–MS/MS analysis

Selected protein bands excised from the SDS/PAGE gel were subjected to in-gel tryptic digestion as described by Wang et al. [21]. For each sample, the peptides were resuspended in 25 μl of 5% (v/v) acetonitrile/0.1% formic acid, of which 8 μl was analysed using a 1100 series capillary LC system (Agilent Technologies, Palo Alto, CA, U.S.A.) coupled directly online with an LCQ Deca ion-trap mass spectrometer [Thermo Finnigan (now Thermo Fisher Scientific), San Jose, CA, U.S.A.]. The instrument was equipped with an in-house manufactured electrospray interface and operated in positive-ion mode with an electrospray voltage of 2.2 kV. The reversed-phase capillary HPLC column containing 5 μm-particle-size Jupiter C18 stationary phase (Phenomenex, Torrance, CA,) was slurry-packed in-house into a 150-μm-internal-diameter×50-cm-long capillary (Polymicro Technologies, Phoenix, AZ, U.S.A.). The mobile phase consisted of (A) 0.1% formic acid in water and (B) 0.1% formic acid in acetonitrile. After loading a sample volume of 8 μl on to the reversed-phase column, the mobile phase was held at 5% B for 20 min and the peptides were eluted using a linear gradient to 95% B for 90 min at a flow rate of 1.5 μl/min. The data acquisition sequence used for all LC–MS/MS analyses employed a survey MS scan followed by four MS/MS scans, where the four most intense ions were dynamically selected from the survey MS scan and subjected to collision-induced dissociation using a normalized collision energy setting of 45%.

Peptides obtained from the pull-down experiments were identified by searching the product ion spectra against the A. thaliana protein database obtained from TAIR (The Arabidopsis Information Resource) website (http://www.arabidopsis.org) using TurboSEQUEST as provided in BioWorks 3.1 (Thermo Finnigan). Peptides obtained from the kinase experiments were identified by searching against a database containing the His-UFD1 protein sequence and the non-redundant E. coli. database obtained from the NCBI (National Center for Biotechnology Information) website (http://www.ncbi.nlm.nih.gov/). Tryptic peptides displaying a charge dependent cross-correlation score (Xcorr) of 2.0, 1.5 and 3.3, for the +1, +2 and +3 charged species of the precursor ions and a delta-correlation score (ΔCn) of 0.08 were used for initial identification [21]. All searches incorporated mass modifications (unified atomic mass unit, u) that included a static carboxamidomethyl modification of 57.0 u on cysteine residues due to alkylation, a differential modification of 16.0 u on methionine residues due to oxidation (methionine sulfoxide) and a differential phosphorylation modification of 80.0 u on serine, threonine and tyrosine residues. The product ion spectra of all phosphopeptides were manually inspected to ensure acceptable ion coverage and phosphorylation-site identification, including those ions involved in neutral losses of HPO3 and H3PO4. Product ion spectra from the pull-down experiments were manually inspected in cases where a protein was identified by a single unique peptide.

RESULTS

Phylogeny and domain organization of the putative type II AtPI4Ks

Phylogenetic analysis derived from the amino acid sequence alignment using the PI3/4K domain from human, yeast and Arabidopsis proteins (InterPro IPR000403) reveals that the PI3/4K-domain-containing proteins fall into three major groups, indicated by the clusters A, B and C in the phylogenetic tree shown in Supplementary Figure S1 at http://www.BiochemJ.org/bj/409/bj4090117add.htm. Cluster A (Supplementary Figure S1) grouped all the PIKKs, which are high-molecular-mass (over 200 kDa) proteins that have serine/threonine kinase activity. In this cluster, human TRRAP (transformation/transcription domain-associated protein) and yeast TRA1 (transcription-associated protein 1) are exceptions in that they do not have kinase activity. The cluster B (Supplementary Figure S1) grouped the canonical PI3Ks and the PI4Ks. This cluster can be separated further into PI3Ks (class I and II), type III PI4Ks and VPS (vacuolar protein sorting)-like PI3K (class III). While the type III PI4Ks and VPS-like PI3K use exclusively inositol phospholipids as substrate, some PI3Ks can use both lipid and protein substrates. For example, the human class Ib PI3Kγ phosphorylates PtdIns(4,5)P2 and non-muscle tropomyosin [22]. Cluster C (Supplementary Figure S1) grouped the human and yeast type II PI4Ks along with a group of eight putative orthologues from Arabidopsis (AtPI4Kγ1–AtPI4Kγ8 according to Mueller-Roeber and Pical [5]). The locus At1G27570 is an exception in the cluster C. Despite its similarity to the type II PI4Ks within the PI3/4K domain, the overall domain organization of the At1G27570 predicted product resembles a typical mem-ber of the PIKK group.

The cluster containing the putative type II PI4K (AtPI4Kγs) from Arabidopsis can be divided in three subgroups. The subdivisions of the AtPI4Kγs extend beyond the sequence similarity within the PI3/4K domain and reflect the presence of UBL domains. AtPI4Kγ2, AtPI4Kγ3 and AtPI4Kγ4 have two N-terminal UBL domains and AtPI4Kγ5, AtPI4Kγ6 and AtPI4Kγ7 have one (Figure 1A). Both AtPI4Kγ1 and AtPI4Kγ8 lack a UBL domain. The linear representations of AtPI4Kγs are shown for comparison with the type II PI4Ks from humans and S. cerevisiae. Figure 1(B) and Table 1 show respective alignments and the percentage of sequence similarity/identity between human and Arabidopsis ubiquitin and the UBL domains from AtPI4Kγ4 and AtPI4Kγ7.

Domain organization of putative type II AtPI4Ks

Figure 1
Domain organization of putative type II AtPI4Ks

(A) The PI3/4K domain is located as indicated (hatched region). In Arabidopsis, two of the three subgroups of putative type II AtPI4Ks have N-terminal UBL domains (black). Type II PI4Ks from humans (HsPI4KII) and S. cerevisiae (ScLSB6) are shown for comparison. (B) Amino acid sequence alignment comparing human (Hs) and Arabidopsis (At) ubiquitin with UBL domains from AtPIKγ4 and AtPIKγ7. The alignment was performed using CLUSTALW (available at http://www.genome.jp/).

Figure 1
Domain organization of putative type II AtPI4Ks

(A) The PI3/4K domain is located as indicated (hatched region). In Arabidopsis, two of the three subgroups of putative type II AtPI4Ks have N-terminal UBL domains (black). Type II PI4Ks from humans (HsPI4KII) and S. cerevisiae (ScLSB6) are shown for comparison. (B) Amino acid sequence alignment comparing human (Hs) and Arabidopsis (At) ubiquitin with UBL domains from AtPIKγ4 and AtPIKγ7. The alignment was performed using CLUSTALW (available at http://www.genome.jp/).

Table 1
Pairwise percentage of similarity/identity of the amino acid sequence between Hs Ub (human ubiquitin, P62988) and At Ub (Arabidopsis ubiquitin, NP_568397) and AtPI4Kγ4 (γ4) and AtPI4Kγ7 (γ7) (UbDKγ4 and UbDKγ7) UBL domains

The sequence alignment was performed using CLUSTALW.

 Percentage of similarity/identity of the amino acid sequence 
 Hs Ub At Ub γ4 UBL1 γ4 UBL2 γ7 UBL 
Hs Ub 100/100 100/96 77/36 68/34 63/23 
At Ub  100/100 78/35 68/34 62/23 
γ4 UBL1   100/100 61/23 68/27 
γ4 UBL2    100/100 52/25 
γ7 UBL     10/100 
 Percentage of similarity/identity of the amino acid sequence 
 Hs Ub At Ub γ4 UBL1 γ4 UBL2 γ7 UBL 
Hs Ub 100/100 100/96 77/36 68/34 63/23 
At Ub  100/100 78/35 68/34 62/23 
γ4 UBL1   100/100 61/23 68/27 
γ4 UBL2    100/100 52/25 
γ7 UBL     10/100 

According to InterPro database and protein searches based on the homology by domain architecture using the Conserved Domain Architecture Retrieval Tool-CDART (http://www.ncbi.nlm.nih.gov/Structure/lexington/lexington.cgi?cmd=rps), proteins displaying the association of the PI3/4K domain with N-terminal UBL domains are not found in metazoans, budding and fission yeasts or in any prokaryotic organism. However, proteins with N-terminal UBL and C-terminal PI3/4K domains are found in Arabidopsis and other dicotyledonous and monocotyledonous plants, algae and protozoans (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/409/bj4090117add.htm), indicating that this novel group of enzymes is conserved not only in the genome of plants, but across genomes of several eukaryotic organisms.

Activity of the AtPI4Kγs

We cloned into expression vectors the coding region of three selected isoforms, AtPI4Kγ1, AtPI4Kγ4 and AtPI4Kγ7, one from each of the three subgroups of the Arabidopsis type II PI4Ks (Figure 1A). Each of the three recombinant proteins was successfully produced in E. coli as an N-terminal GST fusion. Phosphorylation assays carried out in the presence of [γ-32P]ATP and in the absence of any other substrate revealed that recombinant AtPI4Kγ4 and AtPI4Kγ7, but not AtPI4Kγ1, would autophosphorylate (Figure 2A). None of the E. coli-expressed GST-fusion proteins showed detectable PIK activity in in vitro assays in the presence of PtdIns or Type I Folch lipid fraction and [γ-32P] ATP (results not shown).

AtPI4Kγ4 and AtPIKγ7 are protein kinases

Figure 2
AtPI4Kγ4 and AtPIKγ7 are protein kinases

Recombinant GST and GST–AtPI4Kγ1, GST–AtPI4Kγ4 and GST–AtPI4K γ7 produced in E. coli (1 μg) were used in protein phosphorylation reaction mixtures containing [γ-32P]ATP. The reactions were conducted with (A) no protein substrate, (B) 5 μg of type III-S histone or (C) 5 μg of MBP as protein substrates. The products of the reactions were resolved by SDS/PAGE. The 32P-labelled proteins were revealed by autoradiography (upper panels) and the protein amounts by Coomassie Blue staining (lower panels). The arrows in (B) and (C) indicate the substrates used in the assays. Hereafter AtPI4Kγ4 and AtPI4Kγ7 are called UbDKγ4 and UbDKγ7.

Figure 2
AtPI4Kγ4 and AtPIKγ7 are protein kinases

Recombinant GST and GST–AtPI4Kγ1, GST–AtPI4Kγ4 and GST–AtPI4K γ7 produced in E. coli (1 μg) were used in protein phosphorylation reaction mixtures containing [γ-32P]ATP. The reactions were conducted with (A) no protein substrate, (B) 5 μg of type III-S histone or (C) 5 μg of MBP as protein substrates. The products of the reactions were resolved by SDS/PAGE. The 32P-labelled proteins were revealed by autoradiography (upper panels) and the protein amounts by Coomassie Blue staining (lower panels). The arrows in (B) and (C) indicate the substrates used in the assays. Hereafter AtPI4Kγ4 and AtPI4Kγ7 are called UbDKγ4 and UbDKγ7.

Enzymes capable of autophosphorylation often phosphorylate other proteins, so we investigated whether the putative type II AtPI4Kγ4 and AtPI4Kγ7 could also phosphorylate protein substrates. In vitro phosphorylation assays using [γ-32P]ATP and protein substrates revealed that recombinant AtPI4Kγ4 and AtPI4Kγ7, but not AtPI4Kγ1, displayed protein kinase activity. AtPI4Kγ4 and AtPI4Kγ7 phosphorylate type III-S histone (Figure 2B) and MBP (Figure 2C) in vitro. Even though substrate phosphorylation by AtPI4Kγ7 is less evident in comparison with that of AtPI4Kγ4, it can be detected readily using higher amounts of substrates (Supplementary Figure S3 at http://www.BiochemJ.org/bj/409/bj4090117add.htm). We concluded that AtPI4Kγ4 and γ7 are protein kinases and renamed them UbDKγ4 and 'UbDKγ7. Owing to the higher kinase activity of UbDKγ4 compared with that of UbDKγ7, all subsequent experiments to characterize the UbDKs were conducted using UbDKγ4.

Ca2+ plays an important role in stimulating protein phosphorylation by activating Ca2+-regulated protein kinases (for a review, see [23]). In order to determine whether the protein kinase activity of UbDKs is affected by Ca2+, we used the E. coli-expressed UbDKγ4 in a series of in vitro assays where autophosphorylation and phosphorylation of histone were analysed. As shown in Supplementary Figures S4(A) and S4(B) at http://www.BiochemJ.org/bj/409/bj4090117add.htm, Ca2+ does not stimulate either UbDKγ4 autophosphorylation (Supplementary Figure S4A) or type III-S histone phosphorylation (Supplementary Figure S4B). If anything, the presence of 1 mM EGTA caused a slight increase in histone phosphorylation (Supplementary Figure S4B; lane 2 versus lane 3). Interestingly, whereas Ca2+ did not increase UbDKγ4 activity, the presence of histone, a positively charged protein, enhanced autophosphorylation (Supplementary Figure S4C).

LY29400 and wortmannin are known PI3/4K inhibitors. They not only inhibit lipid kinase activity of many PI3Ks and PI4Ks, but also inhibit protein kinase activity of PIKKs [24]. In vitro assays using recombinant UbDKγ4 revealed that, even at high (100 μM) concentrations, these inhibitors had no effect on autophosphorylation or phosphorylation of histone or UFD1 (results not shown).

UBL domains are not necessary for UbDKγ4 autophosphorylation

To determine the role of the different domains in the activity of UbDKs, three N-terminal truncations of UbDKγ4 were generated and produced as GST fusions in E. coli (Figure 3A). The absence of one (ΔUBL1) or two (ΔUBL1/UBL2) UBL domains did not affect UbDKγ4 autophosphorylation (Figure 3B). However, no detectable autophosphorylation was observed in the absence of a 75-amino-acid-long interdomain region between UBL2 and the PI3K/PI4K domain, suggesting that this region is required for autophosphorylation.

UBL domains are not necessary for UbDKγ4 autophosphorylation

Figure 3
UBL domains are not necessary for UbDKγ4 autophosphorylation

(A) UbDKγ4 truncations constructed for expression in E. coli. (B) Recombinant full-length UbDKγ4, all N-terminal truncations (ΔUBL1, ΔUBL1/UBL2 and ΔN) and a C-terminal truncation (ΔC) fused to GST (1 μg each) were assayed for autophosphorylation by adding [γ-32P]ATP as described in the Experimental section. The products of the reactions were resolved by SDS/PAGE. The 32P-labelled proteins were revealed by autoradiography (upper panel) and the protein amounts by Coomassie Blue staining (lower panel).

Figure 3
UBL domains are not necessary for UbDKγ4 autophosphorylation

(A) UbDKγ4 truncations constructed for expression in E. coli. (B) Recombinant full-length UbDKγ4, all N-terminal truncations (ΔUBL1, ΔUBL1/UBL2 and ΔN) and a C-terminal truncation (ΔC) fused to GST (1 μg each) were assayed for autophosphorylation by adding [γ-32P]ATP as described in the Experimental section. The products of the reactions were resolved by SDS/PAGE. The 32P-labelled proteins were revealed by autoradiography (upper panel) and the protein amounts by Coomassie Blue staining (lower panel).

UbDKγ4 is capable of intermolecular autophosphorylation and interaction in vitro

Phosphorylation reactions containing the full-length UbDKγ4 in combination with the inactive N- and C-terminal truncations (ΔN and ΔC respectively; Figure 3A) revealed that the full-length enzyme was able to phosphorylate both protein constructs (Figure 4A). The fact that the ΔN protein is phosphorylated by the full-length enzyme indicates that ΔN is indeed inactive and that its inability to autophosphorylate is not due to absence of a phosphorylation site. The data in Figure 4(A) also indicate that UbDKγ4 has more than one potential site for phosphorylation, i.e., at least one in both the N- and C-terminus of the protein.

UbDKγ4 autophosphorylates multiple sites via intermolecular reaction and interacts with itself

Figure 4
UbDKγ4 autophosphorylates multiple sites via intermolecular reaction and interacts with itself

(A) Recombinant His6–ΔN and His6–ΔC protein constructs (1 μg each) were assayed for phosphorylation in the presence or absence of 1 μg of full-length His6–UbDKγ4 as indicated. (B) Recombinant His6–AtUDPKγ4 (1 μg) was assayed for phosphorylation in the presence of 1 μg of GST or GST–K284A. The products of the reactions were resolved by SDS/PAGE. The 32P-labelled proteins were revealed by autoradiography (upper panels) and the protein amounts by Coomassie Blue staining (lower panels). (C) Glutathione–Sepharose beads, GST or GST–UbDKγ4 immobilized on glutathione–Sepharose beads were used as bait in pull-down assays containing His6–UbDKγ4 as prey. Bound His6–UbDKγ4 was detected by immunoblotting with anti-(His6 tag) antibody (upper panel). GST and GST–UbDKγ4 were visualized by Amido Black staining of the membrane (lower panel). The ‘input’ lane corresponds to 20% of the amount of His6–UbDKγ4 originally used in each binding assay.

Figure 4
UbDKγ4 autophosphorylates multiple sites via intermolecular reaction and interacts with itself

(A) Recombinant His6–ΔN and His6–ΔC protein constructs (1 μg each) were assayed for phosphorylation in the presence or absence of 1 μg of full-length His6–UbDKγ4 as indicated. (B) Recombinant His6–AtUDPKγ4 (1 μg) was assayed for phosphorylation in the presence of 1 μg of GST or GST–K284A. The products of the reactions were resolved by SDS/PAGE. The 32P-labelled proteins were revealed by autoradiography (upper panels) and the protein amounts by Coomassie Blue staining (lower panels). (C) Glutathione–Sepharose beads, GST or GST–UbDKγ4 immobilized on glutathione–Sepharose beads were used as bait in pull-down assays containing His6–UbDKγ4 as prey. Bound His6–UbDKγ4 was detected by immunoblotting with anti-(His6 tag) antibody (upper panel). GST and GST–UbDKγ4 were visualized by Amido Black staining of the membrane (lower panel). The ‘input’ lane corresponds to 20% of the amount of His6–UbDKγ4 originally used in each binding assay.

The ability of full-length UbDKγ4 to trans-phosphorylate ΔN and ΔC protein constructs (Figure 4A) implies that UbDKγ4 may autophosphorylate via an intermolecular reaction. In order to further confirm the intermolecular autophosphorylation, we used versions of full-length UbDKγ4 carrying different N-terminal tags (GST and His6) and a kinase-inactive mutant. The GST- and His6-tagged UbDKγ4 can be easily distinguished by SDS/PAGE, owing due to differences in their molecular masses (∼98 and ∼74 kDa respectively).

To generate a kinase-inactive mutant, we identified the putative ATP-binding site from the N-terminal portion of the PI3K/PI4K domain based on sequence alignments with the catalytic domain of the type II PI4Ks as described by Barylko et al. [25] (results not shown). The alignment predicted that Lys284 of UbDKγ4 was one of the key residues for ATP binding and therefore critical for the enzyme activity. Site-directed mutagenesis was used to change Lys284 to alanine. The mutant, referred here as K284A, was successfully overexpressed in E. coli. Protein-phosphorylation assays comparing wild-type and K284A revealed that the mutation completely abolished protein kinase activity (Supplementary Figure S5 at http://www.BiochemJ.org/bj/409/bj4090117add.htm).

Figure 4(B) shows that kinase-inactive GST–UbDKγ4 (K284A) was phosphorylated by the wild-type His-UbDKγ4. K284A phosphorylation (Figure 4B) was markedly decreased compared with that observed for the ΔN and ΔC protein constructs (Figure 4A). This is probably because the phosphorylation sites of the truncated proteins are more accessible to full-length UbDKγ4. Taken together, these data indicate that UbDKγ4 utilizes an intermolecular reaction mechanism for autophosphorylation, implying at least a transient interaction between two or more UbDKγ4 molecules.

In order to investigate whether UbDKγ4 can establish any kind of stable intermolecular interaction, we performed pull-down experiments using recombinant UbDKγ4 fused with different tags. GST–UbDKγ4 immobilized in glutathione–Sepharose beads was used to pull down His6–UbDKγ4. Immobilized GST–UbDKγ4 bound to His6–UbDKγ4, whereas the controls of unmodified Sepharose beads or immobilized GST elicited no detectable His6–UbDKγ4 binding (Figure 4C). These data indicate that UbDKγ4 can participate in a direct intermolecular interaction.

UbDKγ4 interacts with RPN10 and UFD1 – proteins of the ubiquitin–proteasome system

As a first step toward the functional characterization of the UbDKγs, we attempted to identify putative interacting proteins using pull-down experiments with proteins from Arabidopsis cells growing in suspension culture. Immobilized GST–UbDKγ4 and UbDKγ7 were used to pull down putative interacting proteins. The recovered proteins were resolved by SDS/PAGE, and the most intense Coomassie Blue-stained bands were excised, subjected to in-gel trypsin digestion and the extracted peptides identified by LC–MS/MS analysis, with a list of the best identified candidates appearing in Table 2. On the basis of the literature regarding the function of these proteins and the availability of cDNA clones, seven of these candidates were experimentally tested for direct interaction with UbDKγ4 using recombinant proteins.

Table 2
Putative interacting proteins of UbDKγ4 and UbDKγ7 identified from Arabidopsis using LC–MS/MS analysis

Notes: athe identified peptide sequence may not be unique to one Arabidopsis gene locus owing to the existence of various protein isoforms as determined by performing a BLAST search on each peptide; bannotation according to TAIR website; camino acid residues appearing before and after the full stops correspond to the residues preceding and following the peptide in the protein sequence (on the basis of the product-ion spectra, phosphoserine, phosphothreonine and phosphotyrosine residues are determined as indicated by S*, T* and Y* respectively, and oxidized methionine residues are denoted by M#; daverage mass calculated is based upon the peptide sequence; edeconvoluted mass determined from the observed centroid m/z ratio at the reported charge state; fcharge state of the observed precursor ion; gSEQUEST Xcorr score of the peptide is based on the degree of overlap between the product ion spectrum and the theoretical distribution of b and y ions for the peptide; hSEQUEST ΔCn is the ‘difference’ between the top two Xcorr values for a given product-ion spectrum; idirect protein interactions were determined using in vitro pull-down experiments using recombinant proteins as described in the Experimental section; the interactions indicated as ‘inconclusive’ could not be determined, owing to a failed attempt to produce soluble recombinant protein; the indicated cDNA clones were ordered from the ABRC (http://www.biosci.ohio-state.edu/~plantbio/Facilities/abrc/abrchome.htm/); jpurified endogenous eukaryotic elongation factor 1A from Daucus carota (carrot) cells [45] was used in this experiment. Abbreviations: AGI, Arabidopsis Gene Index; APX3, L-ascorbate peroxidase; Calc., calculated; Chg., charge state; eEF1α, eukaryotic elongation factor 1α; ID#, identification number; Meas., measured; MPPβ, mitochondrial processing peptidase β-subunit precursor; MS5, male sterility MS5 family protein; RPL5B, 60 S ribosomal protein L5; RPN10, component of 26 S proteasome regulatory particle; RPP0C, 60 S acidic ribosomal protein P0; RPS4A, putative ribosomal protein S4.

   [M+H]+     
AGI numbera Proteinb Unique peptides identifiedc Calc. d Meas. e Chg.f Xcorrg ΔCnh Direct interactioni (cDNA ID#) 
Recovered with UbDKγ4         
 At4g38630 RPN10 R.IIVFAGSPIK.Y 1045.3 1044.6 2.105 0.386 Yes (U09430) 
 At3g02090 MPPβ K.LSSDPTTTSQLVANEPASFTGSEVR.M 2595.8 2595.6 3.986 0.464 Inconclusive (U22122) 
 At3g09200 RPP0C K.VEEKEESDEEDYGGDFGLFDEE. 2568.5 2567.6 4.693 0.614 Not determined 
  K.FAVASVAAVSADAGGGAPAAAK.V 1860.1 1859.3 3.653 0.392 (U09559) 
 At5g39740, At3g25520 RPL5B K.M#LEM#DDEYEGNVEATGEDFSVEPTDSR.R 3099.2 3098.1 5.160 0.606 No (U15332 for At3g25520) 
  R.ALLDVGLIR.T 970.2 969.7 2.112 0.317  
 At5g60390, At1g07940 At1g07930, At1g07920 eEF1 α K.NM#ITGTSQADCAVLIIDSTTGGFEAGISK.D 2975.2 2974.0 2.865 0.452 Noj 
  R.LPLQDVYK.I 976.2 976.1 2.597 0.327  
  R.VETGMIKPGMVVTFAPTGLTTEVK.S 2508.0 2509.1 2.543 0.186  
  R.VETGM#IKPGM#VVTFAPTGLTTEVK.S 2540.0 2539.3 2.516 0.329  
  K.IGGIGTVPVGR.V 1026.2 1026.4 2.394 0.278  
  K.M#TPTKPM#VVETFSEYPPLGR.F 2313.7 2312.8 2.214 0.253  
  R.STNLDWYK.G 1027.1 1026.9 2.103 0.296  
  R.STNLDWYK.G 1027.1 1026.5 2.066 0.182  
Recovered with UbDKγ7         
 At2g21270 UFD1 .M#FFDGY*HYHGT*T*FEQSYR.C 2543.4 2542.9 2.163 0.175 Yes 
 At3g02090 MPPβ K.LSSDPTTTSQLVANEPASFTGSEVR.M 2595.8 2595.8 4.383 0.537 Inconclusive (U22122) 
 At2g17360 RPS4A R.LGNVYTIGK.G 965.1 964.5 2.191 0.413 No (U15117) 
  K.FDVGNVVM#VTGGR.N 1367.6 1367.1 2.270 0.323  
 At5g48850 MS5 R.CSKNSQDSLDNVLIDLYKK.C 2241.5 2241.2 2.151 0.086 No (U60768) 
 At4g35000 APX3 K.LSELGFNPNSSAGK.A 1421.5 1420.9 3.478 0.302 No (U16922) 
  K.FDNSYFVELLK.G 1375.6 1375.4 2.433 0.181  
   [M+H]+     
AGI numbera Proteinb Unique peptides identifiedc Calc. d Meas. e Chg.f Xcorrg ΔCnh Direct interactioni (cDNA ID#) 
Recovered with UbDKγ4         
 At4g38630 RPN10 R.IIVFAGSPIK.Y 1045.3 1044.6 2.105 0.386 Yes (U09430) 
 At3g02090 MPPβ K.LSSDPTTTSQLVANEPASFTGSEVR.M 2595.8 2595.6 3.986 0.464 Inconclusive (U22122) 
 At3g09200 RPP0C K.VEEKEESDEEDYGGDFGLFDEE. 2568.5 2567.6 4.693 0.614 Not determined 
  K.FAVASVAAVSADAGGGAPAAAK.V 1860.1 1859.3 3.653 0.392 (U09559) 
 At5g39740, At3g25520 RPL5B K.M#LEM#DDEYEGNVEATGEDFSVEPTDSR.R 3099.2 3098.1 5.160 0.606 No (U15332 for At3g25520) 
  R.ALLDVGLIR.T 970.2 969.7 2.112 0.317  
 At5g60390, At1g07940 At1g07930, At1g07920 eEF1 α K.NM#ITGTSQADCAVLIIDSTTGGFEAGISK.D 2975.2 2974.0 2.865 0.452 Noj 
  R.LPLQDVYK.I 976.2 976.1 2.597 0.327  
  R.VETGMIKPGMVVTFAPTGLTTEVK.S 2508.0 2509.1 2.543 0.186  
  R.VETGM#IKPGM#VVTFAPTGLTTEVK.S 2540.0 2539.3 2.516 0.329  
  K.IGGIGTVPVGR.V 1026.2 1026.4 2.394 0.278  
  K.M#TPTKPM#VVETFSEYPPLGR.F 2313.7 2312.8 2.214 0.253  
  R.STNLDWYK.G 1027.1 1026.9 2.103 0.296  
  R.STNLDWYK.G 1027.1 1026.5 2.066 0.182  
Recovered with UbDKγ7         
 At2g21270 UFD1 .M#FFDGY*HYHGT*T*FEQSYR.C 2543.4 2542.9 2.163 0.175 Yes 
 At3g02090 MPPβ K.LSSDPTTTSQLVANEPASFTGSEVR.M 2595.8 2595.8 4.383 0.537 Inconclusive (U22122) 
 At2g17360 RPS4A R.LGNVYTIGK.G 965.1 964.5 2.191 0.413 No (U15117) 
  K.FDVGNVVM#VTGGR.N 1367.6 1367.1 2.270 0.323  
 At5g48850 MS5 R.CSKNSQDSLDNVLIDLYKK.C 2241.5 2241.2 2.151 0.086 No (U60768) 
 At4g35000 APX3 K.LSELGFNPNSSAGK.A 1421.5 1420.9 3.478 0.302 No (U16922) 
  K.FDNSYFVELLK.G 1375.6 1375.4 2.433 0.181  

Of the seven proteins examined, both RPN10 and UFD1 demonstrated a direct interaction with UbDKγ4. In a reciprocal pull-down assay, immobilized GST–RPN10 and GST–UFD1 were incubated with His6–UbDKγ4. His6–UbDKγ4 was recovered bound to both GST–RPN10 and GST–UFD1, but not with GST alone (Figures 5A and 5B respectively).

UbDKγ4 interacts with RPN10 and UFD1

Figure 5
UbDKγ4 interacts with RPN10 and UFD1

GST, GST–RPN10 (A) and GST–UFD1 (B) immobilized in glutathione–Sepharose beads were used as bait in pull-down assays containing His6–UbDKγ4. The bound UbDKγ4 was visualized by immunoblotting with an anti-(His6 tag) antibody (upper panels). GST, GST-RPN10 and GST-UFD1 were visualized by Amido Black staining of the membrane (lower panels).

Figure 5
UbDKγ4 interacts with RPN10 and UFD1

GST, GST–RPN10 (A) and GST–UFD1 (B) immobilized in glutathione–Sepharose beads were used as bait in pull-down assays containing His6–UbDKγ4. The bound UbDKγ4 was visualized by immunoblotting with an anti-(His6 tag) antibody (upper panels). GST, GST-RPN10 and GST-UFD1 were visualized by Amido Black staining of the membrane (lower panels).

In order to determine which of the UbDKγ4 domains are important for protein–protein interactions, we used the UbDKγ4 truncations (Figure 3A) in binding assays with the recombinant proteins. We were particularly interested to determine which domains were responsible for the UbDKγ4 intermolecular interaction with itself and with RPN10 and UFD1. Because RPN10 and UFD1 bind different forms of ubiquitin, we anticipated that the UBL1 and UBL2 domains of UbDKγ4 would be essential for binding.

Immobilized GST–UbDKγ4 (Figure 6A), GST–RPN10 (Figure 6B) and GST–UFD1 (Figure 6C) were incubated with His6–UbDKγ4, His6–ΔUBL1, His6–ΔN and His6–ΔC. His6–UbDKγ4 was recovered bound to GST–UbDKγ4, GST–RPN10 and GST–UFD1, as expected. His6–ΔUBL1 was detected bound to GST–UbDKγ4 and GST–RPN10, but not to GST–UFD1, indicating that the UBL1 domain was essential for UFD1 binding. Furthermore, the fact that UFD1 bound ΔC and the full-length protein to nearly the same extent, but did not bind ΔUBL1, confirms that the interaction of UFD1 with the UbDKγ4 N-terminal domain (1–257 amino acids) requires the presence of the UBL1 domain (1–107 amino acids). In contrast with UFD1, RPN10 did not bind either ΔN or ΔC. An interaction with RPN10 was detectable with UbDKγ4 and, to a lesser extent, with ΔUBL1. These observations suggest that RPN10 requires sequence component(s) disrupted in both ΔC and ΔN protein constructs, since the absence of the UBL1 domain (ΔUBL1) did not abolish binding.

The N-terminus of UbDKγ4 is important for protein–protein interactions

Figure 6
The N-terminus of UbDKγ4 is important for protein–protein interactions

(A) GST–UbDKγ4, (B) GST–RPN10 and (C) GST–UFD1 immobilized on glutathione–Sepharose beads were used as bait in pull-down assays containing full-length His6–UbDKγ4 and truncations ΔUBL1, ΔN and ΔC as prey. In all binding reactions, immobilized bait and prey were incubated in a 1:2 molar ratio. The bound proteins (indicated by the arrows) were revealed by immunoblotting with anti-(His6 tag) antibody (upper panels). GST-tagged proteins were visualized by Amido Black staining of the membrane (lower panels). The panels in (D) correspond to 20% of the amount of all His6-tagged proteins originally used in each binding assay. The diagram in (E) summarizes the mapping of the UbDKγ4 domains used in the protein–protein interactions. The two non-overlapping regions where RPN10 and UFD1/CDC48 interact and the region required for UbDKγ4 interaction with itself are indicated by the unbroken line. The dotted lines indicate additional regions predicted to participate in these interactions. Data for CDC48 binding are presented in Supplementary Figure S7.

Figure 6
The N-terminus of UbDKγ4 is important for protein–protein interactions

(A) GST–UbDKγ4, (B) GST–RPN10 and (C) GST–UFD1 immobilized on glutathione–Sepharose beads were used as bait in pull-down assays containing full-length His6–UbDKγ4 and truncations ΔUBL1, ΔN and ΔC as prey. In all binding reactions, immobilized bait and prey were incubated in a 1:2 molar ratio. The bound proteins (indicated by the arrows) were revealed by immunoblotting with anti-(His6 tag) antibody (upper panels). GST-tagged proteins were visualized by Amido Black staining of the membrane (lower panels). The panels in (D) correspond to 20% of the amount of all His6-tagged proteins originally used in each binding assay. The diagram in (E) summarizes the mapping of the UbDKγ4 domains used in the protein–protein interactions. The two non-overlapping regions where RPN10 and UFD1/CDC48 interact and the region required for UbDKγ4 interaction with itself are indicated by the unbroken line. The dotted lines indicate additional regions predicted to participate in these interactions. Data for CDC48 binding are presented in Supplementary Figure S7.

For UbDKγ4, the absence of the UBL1 domain (ΔUBL1) decreased, but did not prevent, binding. Additionally, the interaction of UbDKγ4 with the ΔC protein was detectable, but barely. The intermolecular interaction of UbDKγ4 with itself appeared to require a broader sequence region compared with RPN10 and UFD1, suggesting a more extensive self-recognition binding motif that is primarily located within the N-terminal domain.

In summary, component(s) of the UbDKγ4 within the N-terminus, including the UBL domains, were critical for the observed protein–protein interactions that were investigated (Figure 6E). More importantly, RPN10 and UFD1 bind to non-overlapping regions of UbDKγ4. His6–ΔN did not bind to any of the GST-fusion proteins. Interactions with the ΔUBL1/UBL2 protein, although experimentally examined, were not conclusive, because the ΔUBL1/UBL2 protein bound non-specifically to GST (Supplementary Figure S6 at http://www.BiochemJ.org/bj/409/bj4090117add.htm).

In-solution protein structure analysis using NMR revealed that the N-terminal region of UFD1, which is responsible for ubiquitin and polyubiquitin binding, shares a striking structural similarity to an analogous region of CDC48 [26], an abundant AAA-ATPase that forms a homohexamer with chaperone activity, which is necessary for, among other functions, degradation of substrates of the ERAD (endoplasmic-reticulum-associated protein degradation) pathway. The involvement of CDC48 in ERAD depends on the interaction with accessory proteins, such as, UFD2 and the heterodimer UFD1/NPL4 (nuclear protein localization 4) [2731]. To investigate further the binding properties of UbDKγ4, we performed the same binding experiment described above with one of the Arabidopsis CDC48 isoforms, namely CDC48C. GST–CDC48 was used as bait to pull down His6–UbDKγ4 full-length and truncations (see Supplementary Figure S7 at http://www.BiochemJ.org/bj/409/bj4090117add.htm). It was shown that CDC48, like UFD1, can interact with UbDKγ4 in vitro and that this interaction also requires the presence of the UBL1 domain.

UbDKγ4 phosphorylates UFD1 and RPN10

To investigate whether UbDKγ4 would phosphorylate the interacting proteins that we have revealed here, we performed phosphorylation assays in the presence of RPN10 and UFD1 with [γ-32P]ATP. The results of these assays indicated that His6–UbDKγ4 readily phosphorylates GST–UFD1 (Figure 7A), but not GST–RPN10 (Figure 7B). However, RPN10 phosphorylation by UbDKγ4 is detected if the RPN10 substrate is provided as the His6-fusion protein (Figure 7C). UFD1 phosphorylation is always detectable, regardless of the combination of the substrate or enzyme tag (see Supplementary Figure S8 at http://www.BiochemJ.org/bj/409/bj4090117add.htm). Because UDF1 phosphorylation was more robust, we used UFD1 as a substrate to characterize the UbDKγ4 phosphorylation sites.

UbDKγ4 phosphorylates UFD1 and RPN10

Figure 7
UbDKγ4 phosphorylates UFD1 and RPN10

Protein phosphorylation assays were performed with 3 μg of (A) GST–UFD1, (B) GST–RPN10 and (C) 2 μg of His6–RPN10. The reactions were carried in the presence of 2 μg of His6–UbDKγ4 (A and B) or GST–UbDKγ4 (C). In (C), the substrate His6–RPN10 was immobilized on nickel-affinity agarose beads, and GST–UbDKγ4 free in the supernatant was partially removed prior to the electrophoresis. GST–UbDKγ4 inactivation was performed by incubating the enzyme at 90 °C for 10 min. The products of the reactions were resolved by SDS/PAGE. The 32P-labelled proteins were revealed by autoradiography (upper panels) and the protein amounts by Coomassie Blue staining (lower panels).

Figure 7
UbDKγ4 phosphorylates UFD1 and RPN10

Protein phosphorylation assays were performed with 3 μg of (A) GST–UFD1, (B) GST–RPN10 and (C) 2 μg of His6–RPN10. The reactions were carried in the presence of 2 μg of His6–UbDKγ4 (A and B) or GST–UbDKγ4 (C). In (C), the substrate His6–RPN10 was immobilized on nickel-affinity agarose beads, and GST–UbDKγ4 free in the supernatant was partially removed prior to the electrophoresis. GST–UbDKγ4 inactivation was performed by incubating the enzyme at 90 °C for 10 min. The products of the reactions were resolved by SDS/PAGE. The 32P-labelled proteins were revealed by autoradiography (upper panels) and the protein amounts by Coomassie Blue staining (lower panels).

Protein phosphorylation reactions were performed with UbDKγ4 and UFD1 as the substrate, using different combinations of tags. The products of each reaction were separated by SDS/PAGE and the band corresponding to UFD1 was excised and subjected to in-gel tryptic digestion, with the resulting peptides extracted and analysed by LC–MS/MS. With sequence coverage of ∼60%, our analysis revealed that UbDKγ4 targets Ser239/240, Thr294, and Ser311 within the C-terminus of UFD1 (Table 3). No detectable UFD1 phosphorylation was observed for the control reactions performed in the presence of the inactive UbDKγ4 (K284A) or in the absence of ATP. However, it should be noted that, in addition to the identified UFD1 serine/threonine phosphorylation sites, one of the detected phosphoserine residues of the UFD1 construct corresponds to the linker region between the affinity tag and the UFD1 sequence.

Table 3
Identification of UFD1 in vitro phosphorylation sites as determined by LC–MS/MS analysis

Notes: athe amino acid residues appearing before and after the full stops correspond to the residues preceding and following the peptide in the protein sequence (on the basis of the production spectrum, phosphoserine and phosphothreonine residues are determined as indicated by S* and T* respectively; due to the ambiguities in production assignment, only one of the serine residues within the square brackets is phosphorylated; bthe average mass calculated is based upon the peptide sequence; cdeconvoluted mass determined from the observed centroid m/z ratio at the reported charge state; dcharge state of the observed precursor ion; eSEQUEST Xcorr of the peptide is based on the degree of overlap between the production spectrum to the theoretical distribution of b and y ions for the peptide; fSEQUEST ΔCn is the ‘difference’ between the top two Xcorr values for a given production spectrum; gthe total number of b and y ions (identified/theoretical); hamino acid residues containing the phosphorylation site; residue Ser−7 is part of the linker region of the construct that connects the His6 tag to the UFD1 protein and is located seven amino acid residues upstream of the initiator Met1 residue; ithe ΔCn was estimated to be less than 0.1, since the other ranked peptides for the production spectrum elicited lower Xcorr values and did not withstand the rigours of manual interpretation. Abbreviations: Calc., calculated; Chg., charge state; Meas., measured.

 [M+H]+      
Peptidea Calc.b Meas.c Chg.d Xcorre ΔCnf Ionsg Phosphorylation siteh 
K.AGS*AAALFNFK.K 1177.2 1177.7 3.281 0.605 24/30 Ser−7 
R.PLAYEPAPA[SS]*SK.G 1398.4 1397.5 2.484 0.132 19/36 Ser239/Ser240 
K.EAAPKVGAAKET*KKEEQEK.K 2152.3 2150.4 1.536 <0.1i 11/54 Thr294 
K.FQAFS*GKK.Y 992.9 993.0 2.119 0.240 14/21 Ser311 
 [M+H]+      
Peptidea Calc.b Meas.c Chg.d Xcorre ΔCnf Ionsg Phosphorylation siteh 
K.AGS*AAALFNFK.K 1177.2 1177.7 3.281 0.605 24/30 Ser−7 
R.PLAYEPAPA[SS]*SK.G 1398.4 1397.5 2.484 0.132 19/36 Ser239/Ser240 
K.EAAPKVGAAKET*KKEEQEK.K 2152.3 2150.4 1.536 <0.1i 11/54 Thr294 
K.FQAFS*GKK.Y 992.9 993.0 2.119 0.240 14/21 Ser311 

DISCUSSION

Proteins containing the PI3K/PI4K domain include not only predicted lipid kinases, such as the PI4Ks, but also protein kinases such as the members of the PIKK family. Our ongoing objective is to characterize the group of PI3K/PI4K-domain-containing proteins originally identified as putative type II PI4Ks in Arabidopsis [5]. Here we have shown that AtPI4Kγ4 and AtPI4Kγ7, two members of this group which contain the PI3K/PI4K and UBL domains, are protein kinases. In addition, protein–protein interactions revealed that AtPI4Kγ4 interacts with, and phosphorylates in vitro, RPN10 and UFD1, proteins involved in the ubiquitin–proteasome system. We propose that AtPI4Kγ4 and AtPI4Kγ7 are founding members of a new group of protein kinases which we have renamed UbDKs and which belong to the PIKK family of atypical protein kinases [32].

Similarly to other PIKKs, UbDKγ4 and UbDKγ7 autophosphorylate, phosphorylate other proteins and display no detectable lipid kinase activity [1]. Intermolecular interaction and autophosphorylation are also common properties between members of the PIKK family and UbDKγ4. The protein kinase ATM (ataxia-telangiectasia mutated), for example, homodimerizes, and the dissociation of the monomers is stimulated by autophosphorylation, which activates ATM kinase activity [33]. The fact that UbDKγ4 interacts with itself suggests that UbDKγ4, like ATM, may dimerize.

The presence of N-terminal UBL domains (1 or 2) in a subset of the AtPI4Kγs is the most distinctive feature of the AtPI4Kγs. Members of the class I and II PI3K have an rbd (Ras binding domain), PI3K_rbd, that falls into the ubiquitin-domain superfold [34]; however, in contrast with the UBL domain of AtPI4Kγ, PI3K_rbd lacks any sequence similarity to ubiquitin. IKKβ [IκB (inhibitor of nuclear factor κB) kinase β], one of the two kinases that phosphorylate IκB, is the only other known protein kinase that contains a UBL domain [35]. The UBL domain of IKKβ, unlike the ones from some AtPI4Kγs, is located at the central portion of the protein and is essential for protein kinase activity (measured by phosphorylation of the substrate GST–IκBα).

RPN10 and UFD1 directly interact with UbDKγ4, targeting non-overlapping regions of the UbDKγ4 N-terminus where UBL domains are located. RPN10 and UFD1 are distinct proteins, but both bind to some form of ubiquitin and are involved in controlled protein degradation via the ubiquitin–proteasome system (for a review, see [16]). RPN10 is found in eukaryotic cells associated with the 26 S proteasome and free in the cytosol. Its domain organization consists of an N-terminal von Willebrand factor-type A domain and one (in yeast) or two (in higher eukaryotes, including plants) C-terminal ubiquitin-interacting motifs. RPN10 in the 26 S proteasome is located more specifically at the 19 S RP (regulatory particle), where it connects the base and the lid, the two subcomplexes of the 19 S RP.

Arabidopsis RPN10 is encoded by a single gene that can complement yeast Rpn10 deletion mutant and, as predicted, Arabidopsis RPN10 binds to polyubiquitin chains [36]. Arabidopsis plants (rpn10-1) expressing an aberrant form of RPN10 due to a T-DNA insertion accumulate unusually high levels of ubiquitinated proteins, indicating defective protein degradation via the 26 S proteasome [37]. Specifically, rpn10-1 plants fail to degrade the short-lived transcription factor ABA INSENSITIVE-5 to wild-type levels and, as a consequence, the rpn10-1 plants display a range of phenotypes consistent with hypersensity to the plant hormone ABA (abscisic acid) [37].

UFD1 was first identified in a yeast mutant screen for defective degradation of short-lived ubiquitin fusion proteins [38]. The best known role of UFD1 is to form a heterodimeric adaptor of CDC48 with NPL4 [39]. The CDC48–UFD1–NPL4 complex is essential for ERAD, a process that drives misfolded or short-lived proteins from the ER (endoplasmic reticulum) to polyubiquitination and degradation via the 26 S proteasome [27,40]. UFD1 consists of two domains, namely the N-terminal UT3 (1–211) and the C-terminal UT6 (215–307) domains. UFD1 binds to mono- and poly-ubiquitin via non-overlapping regions of its UT3 domain that has a remarkable structural similarity to the CDC48 N-terminal region [26].

The Arabidopsis genome encodes three UFD1 and three CDC48 homologues (CDC48A, CDC48B and CDC48C) but no obvious NPL4. Although UFD1 has not been studied extensively, CDC48 has been implicated in membrane fusion during cytokinesis [41]. Furthermore CDC48 was found to be regulated by PUX1 [plant UBX (ubiquitin regulatory X)-domain-containing protein 1]. PUX1 interacts directly with CDC48A and works as its negative regulator by inactivating CDC48A ATPase activity and stimulating hexamer disassembly [42,43].

The simplest explanation for the interaction of UbDKγ4 with proteins involved in 26 S proteasome substrate delivery is that UbDKγ4 itself is a 26 S proteasome substrate and is directed to degradation. For example, Parkin, a protein responsible for the onset of early-age Parkinson's disease, is an E3 ligase that contains an N-terminal UBL domain. Parkin is a short-lived protein compared with its ΔUBL mutants [44]. However, the notion that UbDKγ4 interacts with two proteins that facilitate substrate presentation to the 26 S proteasome with the sole purpose of UbDKγ4 degradation is conceivable, but not very likely, owing to the fact that it would be rather counterproductive and incompatible with UFD1 function. UFD1 as substrate-delivery mediator seems to operate exclusively in ER-derived 26 S proteasome substrates, and there is no evidence, based on amino acid sequence, that indicates that UbDKγ4 is an ER-associated or ER-resident protein.

An alternative and more attractive explanation is that UbDKγ4 uses RPN10 as a docking site in the proteasome and assists UFD1 to deliver polyubiquitinated proteins targeted for degradation. This scenario is supported by the notion that RPN10 and UFD1 have non-overlapping interaction sites within UbDKγ4. Both RPN10 and UFD1 appeared to bind UbDKγ4 via its N-terminus, where UBL1 and UBL2 are located. However, whereas the N-terminal 107-amino-acid region of UbDKγ4, which includes the UBL1, is necessary for UFD1 interaction, RPN10 targets a broader region, and the absence of UBL1 does not prevent RPN10–UbDKγ4 interaction.

As part of a substrate-delivery mechanism, UbDKγ4 might facilitate protein degradation via the 26 S proteasome in a manner similar to that of the UBA (ubiquitin-associated)/UBL-domain-containing proteins, except that the UbDKγ4 interaction with polyubiquitinated proteins would be indirect via UFD1 alone or possibly in complex with CDC48. Another, non-exclusive, possibility is that UbDKγ4 plays a regulatory role by phosphorylating UFD1 and RPN10 and affects their binding properties and functions. Phosphorylation of the C-terminus of UFD1 most likely affects its affinity for other interacting proteins, but not for polyubiquitinated proteins.

In summary, we have identified a new family of protein kinases and show that they may participate in substrate delivery to the 26 S proteasome via direct interaction with RPN10 and UFD1. In addition, we show that UFD1 and RPN10 can be phosphorylated by UbDKγ4 and have identified in vitro phosphorylation sites on UFD1.

We thank Dr Norma L. Houston (Department of Biochemistry, Christopher S. Bond Life Sciences Center, University of Missouri-Columbia, Columbia, MO, U.S.A.) for the assistance with the phylogenetic analysis. This work was funded in part by the CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico, Brazilian Government) (scholarship to R. M. G.), by National Science Foundation grants MCB-0315869 (to W. F. B.) and MCB-0419819 (to M. B. G.) and by the North Carolina Agricultural Research Service.

Abbreviations

     
  • ABRC

    Arabidopsis Biological Resource Center

  •  
  • ATM

    ataxia-telangiectasia mutated

  •  
  • ER

    endoplasmic reticulum

  •  
  • ERAD

    ER-associated degradation

  •  
  • GST

    glutathione transferase

  •  
  • His6

    hexahistidine

  •  
  • IKK

    IκB (inhibitor of nuclear factor κB) kinase

  •  
  • LC–MS/MS

    liquid chromatography–tandem MS

  •  
  • LSB6

    Las binding protein 6

  •  
  • MBP

    myelin basic protein

  •  
  • NPL4

    nuclear protein localization 4

  •  
  • PI

    phosphoinositide

  •  
  • PIK

    phosphoinositide kinase

  •  
  • PI3K

    phosphoinositide 3-kinase

  •  
  • PI3/4K

    phosphoinositide 3/4-kinase

  •  
  • PI4K

    phosphoinositide 4-kinase

  •  
  • PIKK

    PIK-related kinase

  •  
  • PUX

    plant UBX (ubiquitin regulatory X)-domain-containing protein

  •  
  • rbd

    Ras-binding domain

  •  
  • RP(N)

    regulatory particle (non-ATPase)

  •  
  • RT-PCR

    reverse-transcription PCR

  •  
  • SUMO

    small ubiquitin-related modifier

  •  
  • UBA

    ubiquitin-associated

  •  
  • UbDK

    ubiquitin-like-domain kinase

  •  
  • UBL

    ubiquitin-like

  •  
  • UFD

    ubiquitin fusion degradation

  •  
  • VPS

    vacuolar protein sorting

References

References
1
Abraham
R. T.
PI 3-kinase related kinases: ‘big’ players in stress-induced signaling pathways
DNA Repair
2004
, vol. 
3
 (pg. 
883
-
887
)
2
Balla
A.
Balla
T.
Phosphatidylinositol 4-kinases: old enzymes with emerging functions
Trends Cell Biol.
2006
, vol. 
16
 (pg. 
351
-
361
)
3
Fruman
D. A.
Meyers
R. E.
Cantley
L. C.
Phosphoinositide kinases
Ann. Rev. Biochem.
1998
, vol. 
67
 (pg. 
481
-
507
)
4
Wullschleger
S.
Loewith
R.
Hall
M. N.
TOR signaling in growth and metabolism
Cell
2006
, vol. 
124
 (pg. 
471
-
484
)
5
Mueller-Roeber
B.
Pical
C.
Inositol phospholipid metabolism in Arabidopsis. Characterized and putative isoforms of inositol phospholipid kinase and phosphoinositide-specific phospholipase C
Plant Physiol.
2002
, vol. 
130
 (pg. 
22
-
46
)
6
Stevenson
J. M.
Perera
I. Y.
Boss
W. F.
A phosphatidylinositol 4-kinase pleckstrin homology domain that binds phosphatidylinositol 4-monophosphate
J. Biol. Chem.
1998
, vol. 
273
 (pg. 
22761
-
22767
)
7
Stevenson-Paulik
J.
Love
J.
Boss
W. F.
Differential regulation of two Arabidopsis type III phosphatidylinositol 4-kinase isoforms. A regulatory role for the pleckstrin homology domain
Plant Physiol.
2003
, vol. 
132
 (pg. 
1053
-
1064
)
8
Xue
H.-W.
Pical
C.
Brearley
C.
Elge
S.
Müller-Röber
B.
A plant 126-kDa phosphatidylinositol 4-kinase with a novel repeat structure. Cloning and functional expression in baculovirus-infected insect cells
J. Biol. Chem.
1999
, vol. 
274
 (pg. 
5738
-
5745
)
9
Lou
Y.
Ma
H.
Lin
W.-H.
Chu
Z.-Q.
Mueller-Roeber
B.
Xu
Z.-H.
Xue
H.-W.
The highly charged region of plant β-type phosphatidylinositol 4-kinase is involved in membrane targeting and phospholipid binding
Plant Mol. Biol.
2006
, vol. 
60
 (pg. 
729
-
746
)
10
Han
G. S.
Audhya
A.
Markley
D. J.
Emr
S. D.
Carman
G. M.
The Saccharomyces cerevisiae LSB6 gene encodes phosphatidylinositol 4-kinase activity
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
47709
-
47718
)
11
Shelton
S. N.
Barylko
B.
Binns
D. D.
Horazdovsky
B. F.
Albanesi
J. P.
Goodman
J. M.
Saccharomyces cerevisiae contains a Type II phosphoinositide 4-kinase
Biochem. J.
2003
, vol. 
371
 (pg. 
533
-
540
)
12
Hartmann-Petersen
R.
Gordon
C.
Integral UBL domain proteins: a family of proteasome interacting proteins
Semin. Cell Dev. Biol.
2004
, vol. 
15
 (pg. 
247
-
259
)
13
Larsen
C. N.
Wang
H.
The ubiquitin superfamily: Members, features, and phylogenies
J. Proteome Res.
2002
, vol. 
1
 (pg. 
411
-
419
)
14
Walters
K. J.
Goh
A. M.
Wang
Q.
Wagner
G.
Howley
P. M.
Ubiquitin family proteins and their relationship to the proteasome: a structural perspective
Biochim. Biophys. Acta
2004
, vol. 
1695
 (pg. 
73
-
87
)
15
Hartmann-Petersen
R.
Gordon
C.
Proteins interacting with the 26S proteasome
Cell Mol. Life Sci.
2004
, vol. 
61
 (pg. 
1589
-
1595
)
16
Elsasser
S.
Finley
D.
Delivery of ubiquitinated substrates to protein-unfolding machines
Nat. Cell. Biol.
2005
, vol. 
7
 (pg. 
742
-
749
)
17
Hartmann-Petersen
R.
Seeger
M.
Gordon
C.
Transferring substrates to the 26S proteasome
Trends Biochem. Sci.
2003
, vol. 
28
 (pg. 
26
-
31
)
18
Madura
K.
Rad23 and Rpn10: perennial wallflowers join the melee
Trends Biochem. Sci.
2004
, vol. 
29
 (pg. 
637
-
640
)
19
Yamada
K.
Lim
J.
Dale
J. M.
Chen
H.
Shinn
P.
Palm
C. J.
Southwick
A. M.
Wu
H. C.
Kim
C.
Nguyen
M.
, et al. 
Empirical analysis of transcriptional activity in the Arabidopsis genome
Science.
2003
, vol. 
302
 (pg. 
842
-
846
)
20
Perera
I. Y.
Davis
A. J.
Galanopoulou
D.
Im
Y. J.
Boss
W. F.
Characterization and comparative analysis of Arabidopsis phosphatidylinositol phosphate 5-kinase 10 reveals differences in Arabidopsis and human phosphatidylinositol phosphate kinases
FEBS Lett.
2005
, vol. 
579
 (pg. 
3427
-
3432
)
21
Wang
X.
Goshe
M. B.
Soderblom
E. J.
Phinney
B. S.
Kuchar
J. A.
Li
J.
Asami
T.
Yoshida
S.
Huber
S. C.
Clouse
S. D.
Identification and functional analysis of in vivo phosphorylation sites of the Arabidopsis BRASSINOSTEROID-INSENSITIVE1 receptor kinase
Plant Cell.
2005
, vol. 
17
 (pg. 
1685
-
1703
)
22
Naga Prasad
S. V.
Jayatilleke
A.
Madamanchi
A.
Rockman
H. A.
Protein kinase activity of phosphoinositide 3-kinase regulates β-adrenergic receptor endocytosis
Nat. Cell Biol.
2005
, vol. 
7
 (pg. 
785
-
796
)
23
Harper
J. F.
Breton
G.
Harmon
A.
Decoding Ca2+ signals through plant protein kinases
Annu. Rev. Plant Biol.
2004
, vol. 
55
 (pg. 
263
-
288
)
24
Chiang
G. G.
Abraham
R. T.
Determination of the catalytic activities of mTOR and other members of the phosphoinositide-3-kinase-related kinase family
Methods Mol. Biol.
2004
, vol. 
281
 (pg. 
125
-
141
)
25
Barylko
B.
Wlodarski
P.
Binns
D. D.
Gerber
S. H.
Earnest
S.
Sudhof
T. C.
Grichine
N.
Albanesi
J. P.
Analysis of the catalytic domain of phosphatidylinositol 4-kinase type II
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
44366
-
44375
)
26
Park
S.
Isaacson
R.
Kim
H. T.
Silver
P. A.
Wagner
G.
Ufd1 exhibits the AAA-ATPase fold with two distinct ubiquitin interaction sites
Structure.
2005
, vol. 
13
 (pg. 
995
-
1005
)
27
Bays
N. W.
Hampton
R. Y.
Cdc48–Ufd1–Npl4: stuck in the middle with Ub
Curr Biol.
2002
, vol. 
12
 (pg. 
R366
-
371
)
28
Kostova
Z.
Wolf
D. H.
For whom the bell tolls: protein quality control of the endoplasmic reticulum and the ubiquitin–proteasome connection
EMBO J.
2003
, vol. 
22
 (pg. 
2309
-
2317
)
29
Meyer
H. H.
Wang
Y.
Warren
G.
Direct binding of ubiquitin conjugates by the mammalian p97 adaptor complexes, p47 and Ufd1–Npl4
EMBO J.
2002
, vol. 
21
 (pg. 
5645
-
5652
)
30
Ye
Y.
Meyer
H. H.
Rapoport
T. A.
The AAA ATPase Cdc48/p97 and its partners transport proteins from the ER into the cytosol
Nature
2001
, vol. 
414
 (pg. 
652
-
656
)
31
Ye
Y.
Meyer
H. H.
Rapoport
T. A.
Function of the p97–Ufd1–Npl4 complex in retrotranslocation from the ER to the cytosol: dual recognition of non-ubiquitinated polypeptide segments and polyubiquitin chains
J. Cell Biol.
2003
, vol. 
162
 (pg. 
71
-
84
)
32
Manning
G.
Whyte
D. B.
Martinez
R.
Hunter
T.
Sudarsanam
S.
The protein kinase complement of the human genome
Science
2002
, vol. 
298
 (pg. 
1912
-
1934
)
33
Bakkenist
C. J.
Kastan
M. B.
DNA damage activates ATM through intermolecular autophosphorylation and dimer dissociation
Nature
2003
, vol. 
421
 (pg. 
499
-
506
)
34
Kiel
C.
Serrano
L.
The ubiquitin domain superfold: structure-based sequence alignments and characterization of binding epitopes
J. Mol. Biol.
2006
, vol. 
355
 (pg. 
821
-
844
)
35
May
M. J.
Larsen
S. E.
Shim
J. H.
Madge
L. A.
Ghosh
S.
A novel ubiquitin-like domain in IκB kinase β is required for functional activity of the kinase
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
45528
-
45539
)
36
van Nocker
S.
Deveraux
Q.
Rechsteiner
M.
Vierstra
R. D.
Arabidopsis MBP1 gene encodes a conserved ubiquitin recognition component of the 26S proteasome
Proc. Natl. Acad. Sci. U.S.A.
1996
, vol. 
93
 (pg. 
856
-
860
)
37
Smalle
J.
Kurepa
J.
Yang
P.
Emborg
T. J.
Babiychuk
E.
Kushnir
S.
Vierstra
R. D.
The pleiotropic role of the 26S proteasome subunit RPN10 in Arabidopsis growth and development supports a substrate-specific function in abscisic acid signaling
Plant Cell
2003
, vol. 
15
 (pg. 
965
-
980
)
38
Johnson
E. S.
Ma
P. C. M.
Ota
I. M.
Varshavsky
A.
A proteolytic pathway that recognizes ubiquitin as a degradation signal
J. Biol. Chem.
1995
, vol. 
270
 (pg. 
17442
-
17456
)
39
Meyer
H. H.
Shorter
J. G.
Seemann
J.
Pappin
D.
Warren
G.
A complex of mammalian Ufd1 and Npl4 links the AAA-ATPase, p97, to ubiquitin and nuclear transport pathways
EMBO J.
2000
, vol. 
19
 (pg. 
2181
-
2192
)
40
Jarosch
E.
Taxis
C.
Volkwein
C.
Bordallo
J.
Finley
D.
Wolf
D. H.
Sommer
T.
Protein dislocation from the ER requires polyubiquitination and the AAA-ATPase Cdc48
Nat. Cell Biol.
2002
, vol. 
4
 (pg. 
134
-
139
)
41
Rancour
D. M.
Dickey
C. E.
Park
S.
Bednarek
S. Y.
Characterization of AtCDC48. Evidence for multiple membrane fusion mechanisms at the plane of cell division in plants
Plant Physiol.
2002
, vol. 
130
 (pg. 
1241
-
1253
)
42
Park
S.
Rancour
D. M.
Bednarek
S. Y.
Protein domain-domain interactions and requirements for the negative regulation of Arabidopsis CDC48/p97 by the plant ubiquitin regulatory X (UBX) domain-containing protein, PUX1
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
5217
-
5224
)
43
Rancour
D. M.
Park
S.
Knight
S. D.
Bednarek
S. Y.
Plant UBX domain-containing protein 1, PUX1, regulates the oligomeric structure and activity of Arabidopsis CDC48
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
54264
-
54274
)
44
Finney
N.
Walther
F.
Mantel
P. Y.
Stauffer
D.
Rovelli
G.
Dev
K. K.
The cellular protein level of parkin is regulated by its ubiquitin-like domain
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
16054
-
16058
)
45
Ransom
W. D.
Lao
P.-C.
Gage
D. A.
Boss
W. F.
Phosphoglycerylethanolamine posttranslational modification of plant eukaryotic elongation factor 1α
Plant Physiol.
1998
, vol. 
117
 (pg. 
949
-
960
)

Supplementary data