AMPK (AMP-activated protein kinase)-related kinases regulate cell polarity as well as proliferation and are activated by the LKB1-tumour suppressor kinase. In the present study we demonstrate that the AMPK-related kinases, NUAK1 (AMPK-related kinase 5) and MARK4 (microtubule-affinity-regulating kinase 4), are polyubiquitinated in vivo and interact with the deubiquitinating enzyme USP9X (ubiquitin specific protease-9). Knockdown of USP9X increased polyubiquitination of NUAK1 and MARK4, whereas overexpression of USP9X inhibited ubiquitination. USP9X, catalysed the removal of polyubiquitin chains from wild-type NUAK1, but not from a non-USP9X-binding mutant. Topological analysis revealed that ubiquitin monomers attached to NUAK1 and MARK4 are linked by Lys29 and/or Lys33 rather than the more common Lys48/Lys63. We find that AMPK and other AMPK-related kinases are also polyubiquitinated in cells. We identified non-USP9X-binding mutants of NUAK1 and MARK4 and find that these are hyper-ubiquitinated and not phosphorylated at their T-loop residue targeted by LKB1 when expressed in cells, suggesting that polyubiquitination may inhibit these enzymes. The results of the present study demonstrate that NUAK1 and MARK4 are substrates of USP9X and provide the first evidence that AMPK family kinases are regulated by unusual Lys29/Lys33-linked polyubiquitin chains.
Mutation of the LKB1 protein kinase in humans causes PJS (Peutz–Jeghers syndrome), a disease in which subjects are predisposed to develop benign and malignant tumours . LKB1 forms a complex with the regulatory proteins termed STRAD (STE20-related adaptor) and MO25 (mouse protein 25) and phosphorylates and activates at least 14 protein kinases that are closely related to AMPK (AMP-activated protein kinase) (reviewed in ). The enzymes activated by LKB1 are AMPKα1, AMPKα2, QSK kinase, SIK (salt-induced kinase), QIK (Qin-induced kinase), MARK (microtubule-affinity-regulating kinase) 1, MARK2, MARK3/Par-1A/C-TAK [TGF (transforming growth factor)-β-activated kinase] 1, MARK4, NUAK1/ARK5 (AMPK-related kinase 5), NUAK2/SNARK (SNF1/AMPK-related kinase), BRSK1/SAD-A (brain-specific kinase 1), BRSK2/SAD-B and SNRK (sucrose-non-fermenting-related kinase) . LKB1 activates these enzymes by phosphorylating a conserved threonine residue located within the T-loop of the kinase domain. AMPKα1 and AMPKα2 have been extensively analysed and are activated by LKB1 in response to conditions that lower cellular energy and increase intracellular levels of 5′-AMP. Activation of AMPK enables cells to control their energy resources under situations of stresses . Less is understood about the cellular roles of other LKB1-activated kinases, that are collectively termed AMPK-related kinases . Genetic analysis suggests that MARK isoforms regulate cell polarity [5–7]. The BRSK/SAD enzymes have recently been shown to control axon initiation during neuronal polarization [8–10]. The functions of the other AMPK-related kinases, including isoforms of NUAK, are largely uncharacterized.
Several AMPK-related kinases, including QSK and MARK isoforms, interact with the 14-3-3 adaptor proteins and have been shown to regulate their localization as well as activity [7,11–14]. In addition to 14-3-3 isoforms, other proteins interacted with overexpressed TAP (tandem affinity purification)-tagged AMPK-related kinases in HEK (human embryonic kidney)-293 cells; however, their physiological roles were not explored further [11,13]. One of these was the deubiquitinating enzyme termed USP9X (ubiquitin-specific protease-9), the orthologue of the Drosophila fat facets (dFAF) [11,13]. USP9X is one of the ∼15% of genes encoded on the X chromosome that escapes X-inactivation, a phenomenon that ensures most genes located on this chromosome are expressed at the same levels in male and female cells. The Y chromosome encodes a gene possessing 91% identity in sequence with USP9X, termed USP9Y that is reportedly expressed in male cells .
USP9X is a large 2547-amino-acid-residue enzyme which is widely expressed in all tissues . Apart from a deubiquitinating catalytic domain located between residues 1531 and 1971 belonging to the USP class of deubiquitinating enzymes, and a putative ubiquitin-like domain (residues 873–963) no other obvious domains are present [16,17]. Genetic analysis in Drosophila, suggests that dFAF regulates the polarity and fate of particular cells in the eye and also plays roles in ovary development as well as embryo viability [18,19]. Overexpression of dFAF in neuronal cells of Drosophila led to profound synaptic overgrowth and disruption of synaptic function . Biochemical as well as genetic data suggest that USP9X interacts with, and de-ubiquitinates, the endocytic adaptor protein epsin-1 [21–23]. This may play a role in the receptor-induced endocytosis regulated by the Notch pathway . Others have reported that USP9X de-ubi-quitinates β-catenin , the ras-target AF-6 , survivin  and more recently the E3 ubiquitin ligase Itch .
Post-translational modification of proteins by the small ubiquitin molecule is emerging as a key regulatory mechanism that is beginning to rival phosphorylation in its global importance . It is also becoming clear that ubiquitination controls almost all aspects of protein function not only stability . The regulation of AMPK or AMPK-related kinases by ubiquitination has not previously been investigated. In the present study we provide the first evidence that the NUAK1 and MARK4 kinases are ubiquitinated in vivo and that these enzymes are deubiquitinated by USP9X. Our results suggest that ubiquitination of NUAK1 and MARK4 does not control their stability, but rather may inhibit their phosphorylation and activation by LKB1. Moreover, our results suggest that polyubiquitination of NUAK1 and MARK4 is mediated by an unusual Lys29 and Lys33 linkage, rather than the more common Lys48 or Lys63 linkages.
MATERIALS AND METHODS
Protein G–Sepharose, calmodulin–Sepharose 4B, glutathione–Sepharose, streptavidin–Sepharose, [γ-32P]ATP and ECL (enhanced chemiluminescence) reagent were purchased from Amersham Biosciences. Protease-inhibitor cocktail tablets, and precast SDS/polyacrylamide/BisTris gels were from Invitrogen. Tween 20, rabbit IgG–agarose, HA (haemagglutinin)–agarose resin, dimethyl pimelimidate, NEM (N-ethylmaleimide) and Polybrene were from Sigma. MG-132 was from Calbiochem. Protein desalting spin columns were from Pierce. Lys48-linked tetra-ubiquitin chains and Lys63-linked tetra-ubiquitin chains were from BostonBiochem. Nonidet P40 was from Fluka. Phosphocellulose P81 paper was from Whatman. The hexahistidine-tagged TEV (tobacco etch virus) protease was expressed in Escherichia coli by Elton Zeqiraj (Wellcome Trust Biocentre, University of Dundee, Dundee, Scotland, U.K.) and purified using nickel–agarose affinity chromatography and gel filtration. All peptides were synthesized by Dr Graham Bloomberg (Department of Biochemistry, School of Medical Science, University of Bristol, Bristol, U.K.).
The following antibodies were raised in sheep and affinity-purified on the appropriate antigen: anti-USP9X (residues 2311–2547 of human USP9; used for immunoblotting and immunoprecipitation), anti-NUAK1 (residues 1–661 of human NUAK1; used for immunoblotting and immunoprecipitation), phospho-anti-T-loop MARK3 (residues 204–218 of human MARK3 phosphorylated at Thr211, TVGGKLDTpFCGSPPY; used for immunoblotting), phospho-anti-T-loop NUAK (residues 201–215 of human NUAK2 phosphorylated at Thr209 HQGKFLQTpFCGSPLY; used for immunoblotting), anti-AMPKα1 (residues 344–358 of rat AMPKα1, CTSPPDSFLDDHHLTR) and the anti-GST (glutathione transferase; raised against the GST protein, used for immunoblotting). Mouse monoclonal antibodies recognizing the HA epitope tag (catalogue number 1666606) and monoclonal antibodies recognizing the FLAG epitope tag were purchased from Roche (catalogue number F3165). Rabbit polyclonal anti-ubiquitin antibody was purchased from DakoCytomation (catalogue number Z0458) and secondary antibodies coupled to HRP (horseradish peroxidase) used for immunoblotting were obtained from Pierce.
Tissue culture, transfection, immunoblotting, restriction-enzyme digests, DNA ligations and other recombinant DNA procedures were performed using standard protocols. All mutagenesis was carried out using the QuikChange® site-directed mutagenesis method (Stratagene) using KOD polymerase (Novagen). DNA constructs used for transfection were purified from E. coli DH5α using the Qiagen plasmid Mega or Maxi kit according to the manufacturer's protocol. All DNA constructs were verified by DNA sequencing, which was performed by The Sequencing Service, School of Life Sciences, University of Dundee, Dundee, Scotland, U.K., using DYEnamic ET terminator chemistry (Amersham Biosciences) on Applied Biosystems automated DNA sequencers.
Lysis buffer contained 50 mM Tris/HCl (pH 7.5), 1 mM EGTA, 1 mM EDTA, 1% (w/v) Nonidet P40, 1 mM sodium orthovanadate, 10 mM sodium β-glycerophosphate, 50 mM sodium fluoride, 5 mM sodium pyrophosphate, 0.27 M sucrose, 1 mM DTT (dithiothreitol) and complete proteinase inhibitor cocktail (one tablet/50 ml). To detect ubiquitination in lysates, 5 mM NEM was added to lysis buffer lacking DTT just prior to use. Buffer A contained 50 mM Tris/HCl (pH 7.5), 0.1 mM EGTA, 0.27 M sucrose and 1 mM DTT. Buffer B contained 50 mM Tris/HCl (pH 7.5), 0.15 M NaCl, 0.27 M sucrose, 1% (w/v) Nonidet P40 and 1 mM DTT. Buffer C contained 50mM Tris/HCl (pH 7.5), 0.15 M NaCl, 1 mM MgCl2, 1 mM imidazole, 2 mM CaCl2, 0.27 M sucrose and 1 mM DTT. Buffer D contained 50 mM Tris/HCl (pH 7.5), 20 mM EGTA, 150 mM NaCl and 5 mM DTT. TBS-Tween buffer contained 50 mM Tris/HCl (pH 7.5), 0.15 M NaCl and 0.2% (v/v) Tween 20. Sample buffer was 1×NuPAGE® LDS (lithium dodecyl sulfate) sample buffer (Invitrogen) containing 0.14 M Tris, 2% (w/v) LDS and 10% (v/v) glycerol (final pH 8.5).
The cloning of NUAK1, MARK4 and other AMPK-related kinases have been described previously . In order to clone USP9X (transcript variant 4, accession number EAW59412) an EST (expressed sequence tag) was ordered from Geneservice (IMAGE CLONE 30422528, NCBI accession number CD657533), which covered the coding region of residues 540–2554 (end). The missing N-terminal 2.2 kb of the USP9X cDNA was obtained by RT (reverse transcriptase)–PCR from HeLa mRNA (Stratagene) using the SuperScript III PCR amplification kit (Invitrogen) and the following oligonucleotides: 5-ATGACAGCCACGACTCGTGGCTCT-3 and 5′-CAAGTCATCCATCATATAGGCTCT-3′. The resulting PCR product was subcloned into pSC-A intermediate vector (Stratagene). To create the full-length USP9X cDNA in several mammalian expression vectors, a 5′-BamH1 site and a 3′-Not1 site was incorporated by PCR into the intermediate constructs, then three-way ligations were carried out using BamH1–Spe1 2.2 kb 5′-end and Spe1–Not1 5.4 kb 3′-end fragments. Full-length USP5 (NCBI accession number P45974) was PCR-amplified from IMAGE clone 3506801(NCBI accession number BC005139) and subcloned as a BamH1–Not1 fragment into several expression vectors. Full-length USP7 (NCBI accession number EAW85194) was PCR-amplified from EST dkfzp434f179q2/NCBI accession number AL046721 (Imagenes) and subcloned as a BamH1–Not1 fragment into several expression vectors. CYLD (cylindromatosis; NP_056062) was PCR-amplified from IMAGE clone 4552767 using the GC-rich PCR system (Roche). The resulting fragment was ligated into pCR2.1 (Invitrogen), sequenced to completion and subcloned into the BamH1 site of pFBHTb (Invitrogen) to form pFNHTb CYLD. Ubiquitin (NCBI P62988) was RT–PCR amplified from total RNA (human peripheral blood) using the one-step RT–PCR Superscript III kit (Invitrogen), then cloned into pSC-a (Stratagene) and sequenced to completion. It was subcloned into the BamH1–Not1 sites of pCMVFLAG-1 to form pCMV-FLAG ubiquitin.
Generation of USP9X stable cell lines
HEK-293 cells were cultured in 10-cm-diameter dishes to 50–70% confluence and transfected with 2 μg of pEGFP-C2-TAP construct , encoding human full-length USP9X wild-type or human full-length mutant USP9X(C1559A) using FuGENE™ reagent (Roche) according to the manufacturer's instructions. After 24 h, G418 was added to the medium to a final concentration of 3 mg/ml, and the medium was changed every 24 h maintaining G418. After 14–20 days, individual surviving colonies expressing low levels of GFP (green fluorescent protein) fluorescence were selected and expanded. FACS analysis was also performed to ensure uniform expression of GFP in the selected cell lines and anti-USP9X immunoblotting analysis of lysed cells was also undertaken to ensure that the expressed proteins migrated as a single-molecular-mass species at the expected apparent molecular mass (the isolated GFP–TAP tag adds 50 kDa to the molecular mass of a protein).
Purification of full-length USP9X protein
The purification method was adapted from the previously described TAP-purification protocol . For each purification, 25 15-cm dishes of confluent HEK-293 cell lines stably expressing wild-type USP9X or catalytically inactive USP9X(C1559A) were employed. Cells were washed twice with ice-cold PBS and lysed in 1 ml of ice-cold lysis buffer. The combined lysates were centrifuged at 26000 g for 30 min at 4 °C and the supernatant incubated with 0.2 ml of rabbit IgG–agarose beads for 1 h at 4 °C. The IgG–agarose was washed extensively with lysis buffer containing 0.15 M NaCl, then with several washes in buffer B prior to incubation with 0.250 ml of buffer B containing 0.1 mg of TEV protease. After 3 h at 4 °C ∼70–90% of the TAP-tagged protein had been cleaved from the IgG–agarose and the eluted protein was incubated with 0.1 ml of rabbit calmodulin–Sepharose equilibrated in buffer C. After 1 h at 4 °C, the calmodulin–Sepharose was washed with buffer C. To elute the protein, the calmodulin–Sepharose was then incubated with 0.1 ml of buffer D for 10 min at 4 °C. The eluate was removed from the beads and the elution repeated two or three times. To remove the NaCl present in the buffer containing the eluate protein, the eluates were centrifuged at 1500 g for 1 min at 4 °C in protein desalting spin columns.
Ubiquitinated NUAK1 was employed as a substrate that was obtained by transfecting HEK-293 cells with HA–NUAK1 and FLAG–ubiquitin. At 36 h post-transfection, cell lysates were generated in the presence of 5 mM NEM and for each assay NUAK1 was immunoprecipitated from 1 mg of cell extracts. Assays were set-up in a total volume of 25 μl containing 50 mM Tris/HCl (pH 8.3), 25 mM KCl, 5 mM MgCl2, 5 mM DTT, 0.5 μg of purified USP9X or other deubiquitination enzyme and immunoprecipitated ubiquitinated HA–NUAK1 conjugated to 5 μl of HA–agarose. The same conditions were employed when tetra-ubiquitin was used as the substrate, except that HA–NUAK1 was replaced with 100 ng of Lys48- or Lys63-linked tetra-ubiquitin. Reactions were initiated by the addition of the deubiquitinating enzyme and performed on a vibrating platform at 30 °C. For reaction mixtures containing ubiquitinated HA–NUAK1, reactions were terminated by brief centrifugation to pellet the HA–NUAK1–agarose conjugate and the supernatant was removed and added to LDS-sample buffer. The agarose resin was washed twice in buffer A containing 0.15 M NaCl and twice with buffer A containing no added NaCl then added to an equal volume of 2×LDS-sample buffer. The reactions employing tetra-ubiquitin as substrates were terminated by addition of LDS-sample buffer. Deubiquitination was assessed by subjecting the samples in LDS-sample buffer to immunoblot analysis with anti-ubiquitin antibodies.
Total cell lysate (10–50 μg) or immunoprecipitated samples were heated at 70 °C for 5 min in LDS sample buffer, then subjected to PAGE and electrotransfer on to nitrocellulose membrane. To improve detection of ubiquitinated proteins nitrocellulose membranes were heated at 100 °C for 5 min in Milli-Q water prior to the blocking step . Membranes were blocked in TBS-Tween buffer containing 10% (w/v) dried non-fat skimmed milk powder. The membranes were probed with 1 μg/ml of indicated sheep antibodies or a 1:1000 dilution of commercial antibodies in TBS-Tween buffer containing 5% (w/v) dried non-fat skimmed milk powder for 16 h at 4 °C. Detection was performed using HRP-conjugated secondary antibodies and ECL reagent.
Expression and purification of USP5, USP7 and CYLD
The pGEX expression constructs encoding human full-length wild-type USP5 or USP7 (214–522) were transformed into E. coli BL21 cells. Cultures (1 litre) were grown at 37 °C in Luria broth containing 100 mg/ml ampicillin until the attenuance (D600) was 0.8. Induction of protein expression was carried out by adding 100 μM IPTG (isopropyl β-D-thiogalactoside) and the cells were cultured for a further 16 h at 26 °C. Cells were isolated by centrifugation (3600 g for 30 min at 4 °C), resuspended in 15 ml of ice-cold lysis buffer and lysed in one round of freeze–thawing, followed by sonication to fragment DNA. The lysates were centrifuged at 4 °C for 30 min at 26000 g, and the recombinant proteins were affinity-purified on glutathione–Sepharose and eluted with buffer A containing 20 mM glutathione. Full-length human histidine-tagged CYLD was expressed in insect cells using the baculovirus expression system and purified on Ni2+-nitrilotriacetate–Sepharose as described previously for PDK1 (phosphoinositide-dependent kinase 1) .
Expression and purification of GST–AMPK-related kinases in HEK-293 cells
Typically ten 10-cm-diameter dishes of HEK-293 cells were cultured and each dish was transfected with 10 μg of the pEBG-2T construct encoding wild-type AMPK-related kinase using the polyethylenimine method . The cells were cultured for a further 36 h and lysed in 0.5 ml of ice-cold lysis buffer, the lysates then being pooled and centrifuged at 4 °C for 10 min at 26000 g. The GST-fusion proteins were purified by affinity chromatography on glutathione–Sepharose and eluted in buffer A containing LDS-sample buffer.
Transfection and generation of lysates
Typically ten 10-cm-diameter dishes of HEK-293 cells were cultured and each dish was transfected with a total of 10 μg of the indicated plasmids using the polyethylenimine method . The cells were cultured for a further 36 h and lysed in 0.5 ml of ice-cold lysis buffer either in the absence or presence of NEM. Cell lysates were clarified by centrifugation at 4 °C for 10 min at 26000 g. Rat brain extracts were generated by homogenizing three rat brains in a 10-fold volume of lysis buffer not containing NEM and clarified as above.
Clarified cell lysates (0.1–1 mg) were subjected to immunoprecipitation employing 5 μg of the indicated antibody conjugated to 5 μl of Protein G–Sepharose or, in the case of HA-epitope-tagged proteins, 5 μg of HA antibody covalently conjugated to 5 μl of agarose resin (purchased from Sigma). Immunoprecipitation was undertaken for 1 h at 4 °C on a vibrating platform and the immunoprecipitates were washed twice with 1 ml of lysis buffer containing 0.15 M NaCl and twice with 1 ml of buffer A lacking DTT. The immunoprecipitates were assayed for kinase activity or analysed by immunoblotting.
In vitro kinase assays employing the AMARA peptide
The activity of the immunoprecipitated AMPK-related kinases was quantified by measurement of phosphorylation of the AMARA (AMARAASAAALRRR) peptide substrate . The immunoprecipitated HA–NUAK1 or HA–MARK4 were incubated in a 50 μl mixture containing 50 mM Tris/HCl (pH 7.5), 0.1% (v/v) 2-mercaptoethanol, 10 mM MgCl2, 0.1 mM EGTA, 0.1 mM [γ-32P]ATP (300 c.p.m./pmol) and 200 μM AMARA peptide for 20 min at 30 °C. Incorporation of [32P]phosphate into the peptide substrate was determined by applying 40 μl of the reaction mixture on to P81 phosphocellulose paper and scintillation counting after washing the papers in phosphoric acid as described previously . One unit of activity was defined as that which catalysed the incorporation of 1 nmol of 32P into the substrate. In assays requiring LKB1 activation, the immunoprecipitated NUAK1 or MARK4 proteins were pre-incubated with 5 μg of LKB1–STRAD–MO25 complex kindly provided by Elton Zeqiraj in buffer A containing 10 mM MgCl2 and 0.1 mM ATP in a final volume of 20 ml. Following the LKB1 activation, the kinases were either assayed using the AMARA peptide substrate or subjected to immunoblot analysis.
Lentivirus production and infection
The pLK0.1 lentiviral-based shRNA (short-hairpin RNA) constructs were employed to knockdown USP9X protein levels. We obtained four independent short-hairpin constructs specific for human USP9X from the MISSION™ TRC-HS 1.0 library (Sigma reference no. NM_004652). Only one of the four constructs tested significantly reduced USP9X protein expression and was employed in the present study (number TRC shRNA, TRCN0000007361 5′CCGGGAGAGTTTATTCACTGTCTTACTCGAGTAAGACAGTGAATAAACTCTCTTTTT-3′). We also used a control shRNA construct in the same vector (5′CCTAAGGTTAAGTCGCCCTCGCTCTAGCGAGGGCGACTTAACCTTAGG3′). To generate lentivirus, HEK-293 cells grown on 10-cm-diameter dishes were transfected with a plasmid mix containing 3.5 μg of the shRNA-encoding plasmid, 3.5 μg of pCMV delta R8.2 (packaging plasmid) and 3.5 μg of pCMV-VSV-G (envelope plasmid) using the polyethylenimine method . At 60 h post-transfection, the virus-containing medium of cells was collected and filtered through a 0.2 μm filter. Typically 6–7 ml of viral supernatant was used to infect HEK-293 cells cultured in 10-cm-diameter dishes in the presence of 5 μg/ml polybrene. After 16 h the virus-containing medium was replaced with fresh medium not containing the virus.
Interaction of USP9X with NUAK1 and MARK4
To corroborate findings of previous proteomic screens, which indicated that USP9X interacted with certain AMPK-related kinases [11,13], we overexpressed 13 AMPK-related kinases and tested whether these interacted with endogenous USP9X in HEK-293 cells. Under these conditions, USP9X interacted with NUAK1 and MARK4, but not with the other AMPK-related kinases (Figure 1A). We also found that endogenous NUAK1 and MARK4 were co-immunoprecipitated with endogenous USP9X from rat brain or HEK-293 cells (Figure 1B).
USP9X interacts specifically with NUAK1 and MARK4
Evidence that NUAK1 and MARK4 are ubiquitinated and constitute substrates for USP9X in vivo
To study whether NUAK1 or MARK4 were ubiquitinated in vivo, HEK-293 cells overexpressing NUAK1 or MARK4 were lysed in the absence or presence of NEM in order to inhibit deubiquitinating enzymes and thus prevent deubiquitination of proteins in cell extracts [16,34]. Ubiquitination was assessed following immunoprecipitation and electrophoresis of NUAK1 and MARK4 on a polyacrylamide gel and immunoblot analysis with an anti-ubiquitin antibody. In the presence of NEM, a higher- molecular-mass ubiquitinated species of immunoprecipitated NUAK1 and MARK4 was readily detected that ran as a smear on a polyacrylamide gel, typical of a polyubiquitinated protein (Figure 2A). When cells were lysed in the absence of NEM, much reduced ubiquitination of NUAK1 or MARK4 was observed. Endogenously expressed NUAK1 from HEK-293 cells was also polyubiquitinated (Figure 2B).
USP9X deubiquitinates NUAK1 and MARK4 in vivo
To explore the role of USP9X in controlling polyubiquitination of NUAK1 and MARK4, we generated HEK-293 cells stably overexpressing either full-length wild-type USP9X or catalytically inactive USP9X(C1559A) and investigated how the overexpression of these forms of USP9X affected the ubiquitination of MARK4 and NUAK1. In cells stably overexpressing wild-type USP9X, no significant ubiquitination of NUAK1 or MARK4 was observed (Figure 2C). In cells stably expressing inactive USP9X(C1559A), NUAK1 and MARK4 were ubiquitinated to a higher level than was observed in the control HEK-293 cells not overexpressing USP9X (Figure 2C), suggesting that this mutant acts in a dominant-negative manner, preventing deubiquitination of NUAK1/MARK4 by endogenous USP9X.
We also investigated how reduction of USP9X expression in HEK-293 cells influenced ubiquitination of NUAK1 and MARK4. For these studies we used a USP9X shRNA-encoding lentivirus, derived from the MISSION™ TRC-HS 1.0 library (Sigma), which reproducibly reduced endogenous USP9X protein expression by 80–90%. Under these conditions, both NUAK1 and MARK4 were ubiquitinated to a markedly greater extent than was observed in control cells (Figure 2D).
Full-length USP9X deubiquitinates NUAK1 in vitro
To verify whether USP9X can directly catalyse the deubiquitination of NUAK1, we purified either full-length wild-type USP9X or catalytically inactive USP9X(C1559A) from stable HEK-293 cells expressing these enzymes as described in the Materials and methods section. The large 2547-residue USP9X protein isolated in this manner was not degraded and was the major Coomassie Blue-staining protein on a polyacrylamide gel (Figure 3A). We incubated the purified USP9X with immunoprecipitated ubiquitinated NUAK1 conjugated to an agarose resin. Incubation of immunoprecipitated ubiquitinated wild-type NUAK1 with USP9X, resulted in a significant release to the supernatant of a polyubiquitinated species migrating within a range of ∼20–250 kDa (Figure 3B). There was no evidence of USP9X catalysing the release of mono-ubiquitin from NUAK1 (Figure 3B). In contrast, catalytically inactive USP9X(C1559A), did not induce detectable deubiquitination of immunoprecipitated NUAK1 in parallel experiments (Figure 3B). We also found that other deubiquitinating enzymes tested, namely USP5, USP7 and CYLD, failed to deubiquitinate NUAK1 (Supplementary Figure 1 at http://www.BiochemJ.org/bj/411/bj4110249add.htm) under conditions where they hydrolysed Lys48- or Lys63-linked tetra-ubiquitin (Figure 3C). We also found that full-length USP9X under conditions in which it deubiquitinated NUAK1, failed to hydrolyse Lys48- or Lys63-linked tetra-ubiquitin (Figure 3C).
Evidence that USP9X deubiquitinates NUAK1
Evidence that polyubiquitination of NUAK1 and MARK4 involves atypical Lys29 and Lys33 linkages
Polyubiquitin chains normally result from one of the seven lysine residues on ubiquitin (Lys6, Lys11, Lys27, Lys29, Lys33, Lys48 or Lys63) forming an isopeptide bond with the C-terminal glycine residue of another ubiquitin molecule . To investigate which of the lysine residue(s) on ubiquitin were required for formation of polyubiquitin chains on NUAK1 and MARK4, these enzymes were co-expressed in HEK-293 cells with FLAG-ubiquitin mutants in which all lysine residues except one were changed to arginine (Figure 4A) or in which individual lysine residues were mutated to arginine (Figure 4B). NUAK1 and MARK4 were then immunoprecipitated and immunoblotted with a FLAG antibody to determine whether these were polyubiquitinated with the mutant forms of the FLAG–ubiquitin. Strikingly, these studies revealed that only ubiquitin molecules containing lysine residues at positions 29 and/or 33 were capable of ubiquitinating NUAK1 or MARK4 (Figure 4A). Consistent with this, ubiquitin molecules lacking either or both of these lysine residues were significantly less able to ubiquitinate NUAK1 or MARK4, whereas mutation of the other lysine residues in ubiquitin did not affect ubiquitination (Figure 4B). Lys29/Lys33-linked polyubiquitin chains were directly linked to NUAK1 rather than co-immunoprecipitated-associated proteins, as NUAK1 when co-transfected with either wild-type ubiquitin or Lys29/Lys33-only ubiquitin mutants, was still polyubiquitinated after its immunoprecipitation from a buffer containing 1% (by mass) SDS (Supplementary Figure 2 at http://www.BiochemJ.org/bj/411/bj4110249add.htm). In parallel experiments, TRAF6 (tumour-necrosis-factor-receptor-associated factor 6) a protein regulated by Lys48 and Lys63 polyubiquitination , was only polyubiquitinated in the presence of ubiquitin chains containing Lys48 or Lys63 (Figure 4A) and mutation of both Lys48 and Lys63 were required to prevent its ubiquitination (Figure 4B).
Lys29 and Lys33 linkages are required for ubiquitination of NUAK1 and MARK4
Evidence that AMPK and other AMPK-related kinases are polyubiquitinated in vivo
Using a similar approach we found that several other AMPK-related kinases, including isoforms of MARK and BRSK, were ubiquitinated in cells to various degrees when co-expressed with either wild-type ubiquitin or mutant ubiquitin containing only Lys29 and Lys33 (Figure 5A). MARK3 was ubiquitinated to the same extent as MARK4; however, MARK2 was not significantly ubiquitinated (Figure 5A). Eight other kinases not related to AMPK were not ubiquitinated when co-transfected with the ubiquitin mutant possessing only Lys29 and Lys33 (Figure 5A). We also found that endogenous AMPK immunoprecipitated from fibroblast cells lysed in the presence of NEM was significantly polyubiquitinated (Figure 5B). In the absence of NEM, conditions where AMPK is normally analysed, AMPK was not significantly polyubiquitinated (Figure 5B).
Lys29 and Lys33 linkages are required for ubiquitination of AMPK and several AMPK-related kinases
Identification of non-USP9X interacting forms of NUAK1 and MARK4
To investigate the functional significance of the NUAK1/MARK4 and USP9X interaction we attempted to generate mutants of NUAK1/MARK4 that were unable to bind USP9X. Fragments of NUAK1 were expressed in HEK-293 cells and their ability to bind to endogenous USP9X was analysed (Figure 6A). This revealed that the removal of a region encompassing the C-terminal boundary of the NUAK1 kinase domain (residues 300–310), abolished binding to USP9X. Removal of the equivalent residues on MARK4 also prevented interaction with USP9X (Figure 6B). We next mutated residues within the kinase C-terminal boundary region and studied the effect that this had on USP9X-binding. This revealed that mutation of Trp305 to alanine (Figure 6C) or phenylalanine, valine or isoleucine (Figure 6D) inhibited binding of NUAK1 to USP9X. Sequence alignments indicate that the Trp305 residue is the only residue within the kinase C-terminal boundary region that is conserved in all AMPK family protein kinases (Figure 6E). Mutation of the equivalent tryptophan residue in MARK4 (Trp309) also prevented USP9X interaction (Figure 6F). Residues surrounding the conserved tryptophan residue are less well-conserved between AMPK-related kinases (Figure 6E) and mutations of these residues had variable effects on USP9X binding (Figures 6C and 6F). The C-terminal boundary of NUAK1 may not interact directly with USP9X, as GST-fusions encompassing this region when expressed in HEK-293 cells failed to interact with endogenous USP9X (Supplementary Figure 3 at http://www.BiochemJ.org/bj/411/bj4110249add.htm). Significantly, wild-type purified USP9X protein was unable to catalyse the deubiquitination of the non-USP9X-binding NUAK1(W305F) mutant, indicating that USP9X needs to interact with NUAK1 in order to deubiquitinate it (Figure 3B).
Identification of non-USP9X-binding forms of NUAK1 and MARK4
Evidence that ubiquitination of NUAK1 and MARK4 may inhibit activity and phosphorylation by LKB1
Modulating the degree of ubiquitination of NUAK1 or MARK4 by overexpression or knockdown of USP9X in HEK-293 cells did not markedly affect stability of these enzymes (Figure 2). Moreover, treatment of cells with the proteasome inhibitor MG132 or lysosome inhibitor leupeptin/NH4Cl in the presence of the protein synthesis inhibitor cycloheximide did not significantly alter the cellular levels of NUAK1 or MARK4 (Supplementary Figure 4 at Http://Www.Biochemj.Org/Bj/411/Bj4110249Add.Htm). We next investigated whether ubiquitination of NUAK1 and MARK4 could be correlated with catalytic activity as well as T-loop phosphorylation. As NEM potently inhibited the catalytic activity of NUAK1 and MARK4 (results not shown), cells were lysed in the absence of NEM for experiments in which protein kinase activity was analysed. Strikingly, in contrast with wild-type NUAK1 or MARK4, the non-USP9X-binding NUAK1(W305A) and MARK4(W309A) mutants, were devoid of catalytic activity and were also not phosphorylated at their T-loop residues when expressed in HEK-293 cells (Figures 7A and 7B, top panel). Consistent with the NUAK1(W305A) and MARK4(W309A) mutants being unable to interact with USP9X, these were ubiquitinated to a greater extent than wild-type enzymes when expressed in HEK-293 cells (Figures 7A and 7B, bottom panel). Analysis of other NUAK1 and MARK4 mutants revealed that ubiquitination, activation and T-loop phosphorylation of these enzymes could also be correlated with their ability to bind USP9X (compare Figure 6C with 7A or Figure 6F with 7B). For example, the NUAK1(H303A), MARK4(I310A), MARK4(N311A), MARK4(G313A) and MARK4(Y314A) mutants, which either fail to interact with USP9X or bind weakly (Figure 6), were hyper-ubiquitinated and displayed low activity and markedly reduced T-loop phosphorylation (Figures 7A and 7B). The other mutants of NUAK1 and MARK4 that interact with USP9X to a greater extent were significantly more active and phosphorylated at their T-loop residues (Figures 7A and 7B). In vitro studies demonstrated that the LKB1–STRAD–MO25 complex phosphorylated and activated the NUAK1(W305A) and MARK4(W309A) non-USP9X-binding mutants, showing that the mutations did not affect intrinsic structure and catalytic properties of these enzymes. The four other non-USP9X-binding mutants of MARK4 (Ile310, Asn311, Gly313 and Tyr314), were also phosphorylated and activated by the LKB1–STRAD–MO25 complex. Only the NUAK1(H303A) mutant was not phosphorylated or activated by LKB1, indicating that this mutation disrupted the structure of the enzyme. However, it should be noted that our results do not rule out the possibility that mutation of Trp305/Trp309 residues destabilizes the structure of NUAK1 and MARK4, in a manner that prevents in vivo phosphorylation by LKB1 independently from ubiquitination and/or its ability to bind USP9X.
Evidence that ubiquitination of NUAK1 and MARK4 inhibits activity and phosphorylation of these enzymes by LKB1
To our knowledge, regulation of AMPK or AMPK-related kinases by ubiquitination has not previously been investigated. It is likely that ubiquitination has been missed in previous studies, as these enzymes are normally analysed after their isolation from cell extracts generated in the absence of inhibitors of deubiquitinating enzymes such as NEM. Our results also indicate that NUAK1 and MARK4 are likely to be deubiquitinated in vivo by USP9X, as overexpression of USP9X inhibited NUAK1 and MARK4 ubiquitination, whereas knockdown of USP9X expression enhanced ubiquitination. Interaction of USP9X with NUAK1 and MARK4 is required for deubiquitination, as the non-USP9X-binding mutants are not substrates for the purified USP9X protein in vitro and these mutants are also hyper-ubiquitinated when expressed in cells. In vitro we were able to demonstrate that full-length USP9X protein catalysed the release of polyubiquitin species from ubiquitinated NUAK1 immunoprecipitated from HEK-293 cells. Under the same conditions, other deubiquitinating enzymes including USP5, USP7 and CYLD failed to deubiquitinate NUAK1. Interaction of USP9X with NUAK1 was required for the deubiquitination, as the non-binding NUAK1(W305A) mutant could not be deubiquitinated by USP9X. We also found that USP9X catalysed the release of polyubiquitinated products from NUAK1 rather than free ubiquitin monomers (Figure 3). This suggests that USP9X interacts with NUAK1 and catalyses release of polyubiquitin chains rather than ubiquitin monomers from the end of the polyubiquitin chain. Overexpression of USP9X in cells prevented detectable ubiquitination of NUAK1 (Figure 2C), indicating that USP9X plays a rate-limiting role in the removal of the entire polyubiquitin chains from NUAK1. Consistent with the notion that USP9X needs to interact with its substrate prior to catalysing deubiquitination, we found that full-length USP9X was unable to hydrolyse Lys48- or Lys63-linked tetra-ubiquitin under conditions in which USP5, USP7 and CYLD acted upon these substrates. Other proposed substrates for USP9X, including epsin-1 [21–23] and Itch , have also been reported to bind USP9X. It would be important to define the region on USP9X that interacts with NUAK1 and MARK4 and determine whether this was also involved in enabling USP9X to associate with its other substrates.
Our results indicate that the polyubiquitin chains attached to NUAK1 and MARK4 are linked through Lys29 and Lys33. Global MS analysis of ubiquitinated proteins has revealed that all lysine residues in ubiquitin, including Lys29 and Lys33, are employed to form branch points [36,37]. Another USP9X substrate, the Itch ligase, was reported to auto-ubiquitinate through Lys29-conjugated chains . As three USP9X substrates, namely NUAK1, MARK4 and Itch, contain Lys29-linked ubiquitin chains, it is tempting to speculate that USP9X may have evolved specificity and/or preference for deubiquitinating proteins conjugated to polyubiquitin chains possessing Lys29 and/or Lys33 linkages. However, it should be noted, in the case of the survivin substrate, that USP9X reportedly deubiquitinates Lys63-linked polyubiquitin chains . Similar to NUAK1/MARK4, USP9X does not affect the stability of survivin, but instead regulates chromosome alignment and segregation by controlling association of survivin with centromeres as well as co-localizing survivin and aurora B to centromeres.
Deubiquitination of Itch by USP9X reportedly protected it from degradation by the lysosome. Similarly, USP9X has been proposed to stabilize the expression of its other substrates epsin-1 , β-catenin [25,39] and AF-6 . In the case of NUAK1 and MARK4, our results indicate that polyubiquitination of these enzymes does not control their stability. Instead the data suggest that interaction of NUAK1 and MARK4 with USP9X regulates their phosphorylation and activation by the LKB1 tumour suppressor (Figure 7). In future work it would be important to identify the ubiquitin ligases that introduce the Lys29/Lys33-linked polyubiquitin chains on to NUAK1 and MARK4. We have also found that several other AMPK-related kinases become ubiquitinated when co-expressed with ubiquitin mutants containing only Lys29 and Lys33 (Figure 5), suggesting that the ubiquitin ligases acting upon NUAK1 and MARK4 might also ubiquitinate other members of the AMPK family. This modification appears specific to AMPK family kinases, as other kinases tested were not ubiquitinated with mutant ubiquitin containing only Lys29 and Lys33 (Figure 5). It will also be important to establish the lysine residues in NUAK1 and MARK4 that the polyubiquitin chains are attached to as well as the role(s) that Lys29 and Lys33-linked polyubiquitination plays and whether these types of chains are capable of interacting with a specific subset of proteins. Endogenous AMPK was also observed to be polyubiquitinated when immunoprecipitated from cells lysed in the presence of NEM. Stimulation of cells with agonists that activate AMPK [AICAR (5-amino-4-imidazolecarboxamide riboside), phenformin and A769662] increased phosphorylation of AMPK at Thr172 (the site of LKB1 phosphorylation), but did not markedly alter ubiquitination status (results not shown). Further work is required to analyse which AMPK subunits are ubiquitinated, what types of ubiquitin chains are attached and how this affects AMPK function.
MARK4, possesses a UBA (ubiquitin-associated domain) located between residues 324 and 368, in close proximity to the region required for interaction with USP9X (residues 300–310). The role of the UBA on MARK4 and several other AMPK-related kinases is poorly understood. Mutations of conserved residues within the UBA of several AMPK-related kinases prevented phosphorylation and activation by LKB1, suggesting that the UBA might regulate the conformation of the kinase domain and its accessibility to LKB1 . Recently, an elegant study demonstrated that the UBA of MARK3 possessed significant intrinsic conformational instability and had a tendency to unfold explaining the very low affinity of this domain for ubiquitin . This study also established that, at the expense of its capacity to engage ubiquitin, the UBA of MARK3 preferentially interacted with the N-terminal lobe of the MARK3 kinase domain, thereby stabilizing it in an open active conformation . An intriguing possibility is that when MARK isoforms are polyubiquitinated, the ubiquitin chains might compete with the kinase domain for the UBA binding. Polyubiquitination of MARK isoforms would therefore act to disrupt the kinase–UBA interaction. This would lead to destabilization of the kinase domain similarly to the UBA mutations that have previously been investigated , and in so doing inhibit LKB1 T-loop phosphorylation. A model of how polyubiquitination might regulate the activity of AMPK-related kinases in presented in Figure 8. Although domain identification programs fail to recognize a UBA in NUAK1, sequence alignments imply some similarity between the UBA on other AMPK-related kinases and the equivalent region on NUAK1, suggesting that NUAK1 may possess a UBA-like motif . The catalytic subunit of AMPK may also contain a UBA-like motif in the equivalent region . Other possibilities of how polyubiquitination of MARK4/NUAK1 might regulate their activity are if polyubiquitin chains sterically shield the T-loop threonine residue and/or a site on these enzymes that interacts with LKB1. Ubiquitination of NUAK1 and MARK4 could also induce conformational changes promoting their dephosphorylation by a protein phosphatase. A major challenge in this area of research is to develop a cell lysis condition that would preserve NUAK1 and MARK4 ubiquitination as well as kinase catalytic activity. In the present study, cells were lysed in the presence of 5 mM NEM in order to maintain ubiquitination, a condition that completely inactivates NUAK1 and MARK4 protein kinase activity. We found that other USP-inhibiting alkylating agents, such as iodoacetimide, also inactivated NUAK1 and MARK4 (results not shown). We also attempted to lyse cells in the presence of the cysteine protease inhibitor E64 as well as ubiquitin aldehyde that do not inhibit kinase activity of NUAK1 and MARK4, but found that these compounds were much less effective than NEM in preserving polyubiquitination of NUAK1/MARK4 (results not shown). Further work is required to develop a potent USP inhibitor that works in cell extracts, but does not inhibit protein kinases.
Model of how USP9X and ubiquitination controls AMPK family kinases
In conclusion, the results of the present study provide the first indication that the family of AMPK kinases is regulated by ubiquitination. They demonstrate that ubiquitinated forms of NUAK1 and MARK4 are substrates for the USP9X deubiquitinating enzyme. Our results indicate that ubiquitination of NUAK1 and MARK4 may regulate their phosphorylation and activation by LKB1, but further work is required to validate this conclusion. More work is also required to define the mechanism by which ubiquitination of NUAK1 and MARK4 regulates these enzymes. It will also be important to define how USP9X is regulated and the significance of the atypical Lys29/Lys33-linked ubiquitin chains attached to NUAK1 and MARK4. It will also be interesting to determine the identity of the ubiquitin ligases that introduce the Lys29- and Lys33-conjugated ubiquitin chains to NUAK1 and MARK4.
We thank the Sequencing Service (School of Life Sciences, University of Dundee, Dundee, Scotland, U.K.) for DNA sequencing, the Post Genomics and Molecular Interactions Centre for Mass Spectrometry facilities (School of Life Sciences, University of Dundee, Dundee, Scotland, U.K.) and the antibody purification teams [Division of Signal Transduction Therapy (DSTT), University of Dundee, Dundee, Scotland, U.K.] co-ordinated by Hilary McLauchlan and James Hastie for expression and purification of antibodies and Grahame Hardie for provision of the AMPKα1 antibody. A. K. A. was supported by a Moffat Charitable Trust studentship and A. Z. by a Wellcome Trust Studentship. We thank the Association for International Cancer Research, Diabetes UK, the Medical Research Council, the Moffat Charitable Trust and the pharmaceutical companies supporting the Division of Signal Transduction Therapy Unit (AstraZeneca, Boehringer-Ingelheim, GlaxoSmithKline, Merck & Co. Inc, Merck KgaA and Pfizer) for financial support.
AMP-activated protein kinase
expressed sequence tag
green fluorescent protein
human embryonic kidney
lithium dodecyl sulfate
mouse protein 25
AMPK-related kinase 5
tandem affinity purification
tumour-necrosis-factor-receptor-associated factor 6